]> Git Repo - linux.git/log
linux.git
6 months agoMerge tag 'regulator-fix-v6.11-stub' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 5 Sep 2024 23:41:16 +0000 (16:41 -0700)]
Merge tag 'regulator-fix-v6.11-stub' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator

Pull regulator fix from Mark Brown:
 "A fix from Doug Anderson for a missing stub, required to fix the build
  for some newly added users of devm_regulator_bulk_get_const() in
  !REGULATOR configurations"

* tag 'regulator-fix-v6.11-stub' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator:
  regulator: core: Stub devm_regulator_bulk_get_const() if !CONFIG_REGULATOR

6 months agoMerge tag 'rust-fixes-6.11-2' of https://github.com/Rust-for-Linux/linux
Linus Torvalds [Thu, 5 Sep 2024 23:35:57 +0000 (16:35 -0700)]
Merge tag 'rust-fixes-6.11-2' of https://github.com/Rust-for-Linux/linux

Pull Rust fixes from Miguel Ojeda:
 "Toolchain and infrastructure:

   - Fix builds for nightly compiler users now that 'new_uninit' was
     split into new features by using an alternative approach for the
     code that used what is now called the 'box_uninit_write' feature

   - Allow the 'stable_features' lint to preempt upcoming warnings about
     them, since soon there will be unstable features that will become
     stable in nightly compilers

   - Export bss symbols too

  'kernel' crate:

   - 'block' module: fix wrong usage of lockdep API

  'macros' crate:

   - Provide correct provenance when constructing 'THIS_MODULE'

  Documentation:

   - Remove unintended indentation (blockquotes) in generated output

   - Fix a couple typos

  MAINTAINERS:

   - Remove Wedson as Rust maintainer

   - Update Andreas' email"

* tag 'rust-fixes-6.11-2' of https://github.com/Rust-for-Linux/linux:
  MAINTAINERS: update Andreas Hindborg's email address
  MAINTAINERS: Remove Wedson as Rust maintainer
  rust: macros: provide correct provenance when constructing THIS_MODULE
  rust: allow `stable_features` lint
  docs: rust: remove unintended blockquote in Quick Start
  rust: alloc: eschew `Box<MaybeUninit<T>>::write`
  rust: kernel: fix typos in code comments
  docs: rust: remove unintended blockquote in Coding Guidelines
  rust: block: fix wrong usage of lockdep API
  rust: kbuild: fix export of bss symbols

6 months agoMerge tag 'trace-v6.11-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/trace...
Linus Torvalds [Thu, 5 Sep 2024 23:29:41 +0000 (16:29 -0700)]
Merge tag 'trace-v6.11-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace

Pull tracing fixes from Steven Rostedt:

 - Fix adding a new fgraph callback after function graph tracing has
   already started.

   If the new caller does not initialize its hash before registering the
   fgraph_ops, it can cause a NULL pointer dereference. Fix this by
   adding a new parameter to ftrace_graph_enable_direct() passing in the
   newly added gops directly and not rely on using the fgraph_array[],
   as entries in the fgraph_array[] must be initialized.

   Assign the new gops to the fgraph_array[] after it goes through
   ftrace_startup_subops() as that will properly initialize the
   gops->ops and initialize its hashes.

 - Fix a memory leak in fgraph storage memory test.

   If the "multiple fgraph storage on a function" boot up selftest fails
   in the registering of the function graph tracer, it will not free the
   memory it allocated for the filter. Break the loop up into two where
   it allocates the filters first and then registers the functions where
   any errors will do the appropriate clean ups.

 - Only clear the timerlat timers if it has an associated kthread.

   In the rtla tool that uses timerlat, if it was killed just as it was
   shutting down, the signals can free the kthread and the timer. But
   the closing of the timerlat files could cause the hrtimer_cancel() to
   be called on the already freed timer. As the kthread variable is is
   set to NULL when the kthreads are stopped and the timers are freed it
   can be used to know not to call hrtimer_cancel() on the timer if the
   kthread variable is NULL.

 - Use a cpumask to keep track of osnoise/timerlat kthreads

   The timerlat tracer can use user space threads for its analysis. With
   the killing of the rtla tool, the kernel can get confused between if
   it is using a user space thread to analyze or one of its own kernel
   threads. When this confusion happens, kthread_stop() can be called on
   a user space thread and bad things happen. As the kernel threads are
   per-cpu, a bitmask can be used to know when a kernel thread is used
   or when a user space thread is used.

 - Add missing interface_lock to osnoise/timerlat stop_kthread()

   The stop_kthread() function in osnoise/timerlat clears the osnoise
   kthread variable, and if it was a user space thread does a put_task
   on it. But this can race with the closing of the timerlat files that
   also does a put_task on the kthread, and if the race happens the task
   will have put_task called on it twice and oops.

 - Add cond_resched() to the tracing_iter_reset() loop.

   The latency tracers keep writing to the ring buffer without resetting
   when it issues a new "start" event (like interrupts being disabled).
   When reading the buffer with an iterator, the tracing_iter_reset()
   sets its pointer to that start event by walking through all the
   events in the buffer until it gets to the time stamp of the start
   event. In the case of a very large buffer, the loop that looks for
   the start event has been reported taking a very long time with a non
   preempt kernel that it can trigger a soft lock up warning. Add a
   cond_resched() into that loop to make sure that doesn't happen.

 - Use list_del_rcu() for eventfs ei->list variable

   It was reported that running loops of creating and deleting kprobe
   events could cause a crash due to the eventfs list iteration hitting
   a LIST_POISON variable. This is because the list is protected by SRCU
   but when an item is deleted from the list, it was using list_del()
   which poisons the "next" pointer. This is what list_del_rcu() was to
   prevent.

* tag 'trace-v6.11-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  tracing/timerlat: Add interface_lock around clearing of kthread in stop_kthread()
  tracing/timerlat: Only clear timer if a kthread exists
  tracing/osnoise: Use a cpumask to know what threads are kthreads
  eventfs: Use list_del_rcu() for SRCU protected list variable
  tracing: Avoid possible softlockup in tracing_iter_reset()
  tracing: Fix memory leak in fgraph storage selftest
  tracing: fgraph: Fix to add new fgraph_ops to array after ftrace_startup_subops()

6 months agoMerge tag 'platform-drivers-x86-v6.11-6' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 5 Sep 2024 16:57:50 +0000 (09:57 -0700)]
Merge tag 'platform-drivers-x86-v6.11-6' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86

Pull x86 platform driver fixes from Ilpo Järvinen:

 - amd/pmf: ASUS GA403 quirk matching tweak

 - dell-smbios: Fix to the init function rollback path

* tag 'platform-drivers-x86-v6.11-6' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86:
  platform/x86/amd: pmf: Make ASUS GA403 quirk generic
  platform/x86: dell-smbios: Fix error path in dell_smbios_init()

6 months agoMerge tag 'linux_kselftest-kunit-fixes-6.11-rc7' of git://git.kernel.org/pub/scm...
Linus Torvalds [Thu, 5 Sep 2024 16:43:38 +0000 (09:43 -0700)]
Merge tag 'linux_kselftest-kunit-fixes-6.11-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest

Pull kunit fix fromShuah Khan:
 "One single fix to a use-after-free bug resulting from
  kunit_driver_create() failing to copy the driver name leaving it on
  the stack or freeing it"

* tag 'linux_kselftest-kunit-fixes-6.11-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest:
  kunit: Device wrappers should also manage driver name

6 months agotracing/timerlat: Add interface_lock around clearing of kthread in stop_kthread()
Steven Rostedt [Thu, 5 Sep 2024 15:33:59 +0000 (11:33 -0400)]
tracing/timerlat: Add interface_lock around clearing of kthread in stop_kthread()

The timerlat interface will get and put the task that is part of the
"kthread" field of the osn_var to keep it around until all references are
released. But here's a race in the "stop_kthread()" code that will call
put_task_struct() on the kthread if it is not a kernel thread. This can
race with the releasing of the references to that task struct and the
put_task_struct() can be called twice when it should have been called just
once.

Take the interface_lock() in stop_kthread() to synchronize this change.
But to do so, the function stop_per_cpu_kthreads() needs to change the
loop from for_each_online_cpu() to for_each_possible_cpu() and remove the
cpu_read_lock(), as the interface_lock can not be taken while the cpu
locks are held. The only side effect of this change is that it may do some
extra work, as the per_cpu variables of the offline CPUs would not be set
anyway, and would simply be skipped in the loop.

Remove unneeded "return;" in stop_kthread().

Cc: [email protected]
Cc: Masami Hiramatsu <[email protected]>
Cc: Mathieu Desnoyers <[email protected]>
Cc: Tomas Glozar <[email protected]>
Cc: John Kacur <[email protected]>
Cc: "Luis Claudio R. Goncalves" <[email protected]>
Link: https://lore.kernel.org/[email protected]
Fixes: e88ed227f639e ("tracing/timerlat: Add user-space interface")
Signed-off-by: Steven Rostedt (Google) <[email protected]>
6 months agotracing/timerlat: Only clear timer if a kthread exists
Steven Rostedt [Thu, 5 Sep 2024 12:53:30 +0000 (08:53 -0400)]
tracing/timerlat: Only clear timer if a kthread exists

The timerlat tracer can use user space threads to check for osnoise and
timer latency. If the program using this is killed via a SIGTERM, the
threads are shutdown one at a time and another tracing instance can start
up resetting the threads before they are fully closed. That causes the
hrtimer assigned to the kthread to be shutdown and freed twice when the
dying thread finally closes the file descriptors, causing a use-after-free
bug.

Only cancel the hrtimer if the associated thread is still around. Also add
the interface_lock around the resetting of the tlat_var->kthread.

Note, this is just a quick fix that can be backported to stable. A real
fix is to have a better synchronization between the shutdown of old
threads and the starting of new ones.

Link: https://lore.kernel.org/all/[email protected]/
Cc: [email protected]
Cc: Masami Hiramatsu <[email protected]>
Cc: Mathieu Desnoyers <[email protected]>
Cc: "Luis Claudio R. Goncalves" <[email protected]>
Link: https://lore.kernel.org/[email protected]
Fixes: e88ed227f639e ("tracing/timerlat: Add user-space interface")
Reported-by: Tomas Glozar <[email protected]>
Signed-off-by: Steven Rostedt (Google) <[email protected]>
6 months agotracing/osnoise: Use a cpumask to know what threads are kthreads
Steven Rostedt [Wed, 4 Sep 2024 14:34:28 +0000 (10:34 -0400)]
tracing/osnoise: Use a cpumask to know what threads are kthreads

The start_kthread() and stop_thread() code was not always called with the
interface_lock held. This means that the kthread variable could be
unexpectedly changed causing the kthread_stop() to be called on it when it
should not have been, leading to:

 while true; do
   rtla timerlat top -u -q & PID=$!;
   sleep 5;
   kill -INT $PID;
   sleep 0.001;
   kill -TERM $PID;
   wait $PID;
  done

Causing the following OOPS:

 Oops: general protection fault, probably for non-canonical address 0xdffffc0000000002: 0000 [#1] PREEMPT SMP KASAN PTI
 KASAN: null-ptr-deref in range [0x0000000000000010-0x0000000000000017]
 CPU: 5 UID: 0 PID: 885 Comm: timerlatu/5 Not tainted 6.11.0-rc4-test-00002-gbc754cc76d1b-dirty #125 a533010b71dab205ad2f507188ce8c82203b0254
 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-debian-1.16.3-2 04/01/2014
 RIP: 0010:hrtimer_active+0x58/0x300
 Code: 48 c1 ee 03 41 54 48 01 d1 48 01 d6 55 53 48 83 ec 20 80 39 00 0f 85 30 02 00 00 49 8b 6f 30 4c 8d 75 10 4c 89 f0 48 c1 e8 03 <0f> b6 3c 10 4c 89 f0 83 e0 07 83 c0 03 40 38 f8 7c 09 40 84 ff 0f
 RSP: 0018:ffff88811d97f940 EFLAGS: 00010202
 RAX: 0000000000000002 RBX: ffff88823c6b5b28 RCX: ffffed10478d6b6b
 RDX: dffffc0000000000 RSI: ffffed10478d6b6c RDI: ffff88823c6b5b28
 RBP: 0000000000000000 R08: ffff88823c6b5b58 R09: ffff88823c6b5b60
 R10: ffff88811d97f957 R11: 0000000000000010 R12: 00000000000a801d
 R13: ffff88810d8b35d8 R14: 0000000000000010 R15: ffff88823c6b5b28
 FS:  0000000000000000(0000) GS:ffff88823c680000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 0000561858ad7258 CR3: 000000007729e001 CR4: 0000000000170ef0
 Call Trace:
  <TASK>
  ? die_addr+0x40/0xa0
  ? exc_general_protection+0x154/0x230
  ? asm_exc_general_protection+0x26/0x30
  ? hrtimer_active+0x58/0x300
  ? __pfx_mutex_lock+0x10/0x10
  ? __pfx_locks_remove_file+0x10/0x10
  hrtimer_cancel+0x15/0x40
  timerlat_fd_release+0x8e/0x1f0
  ? security_file_release+0x43/0x80
  __fput+0x372/0xb10
  task_work_run+0x11e/0x1f0
  ? _raw_spin_lock+0x85/0xe0
  ? __pfx_task_work_run+0x10/0x10
  ? poison_slab_object+0x109/0x170
  ? do_exit+0x7a0/0x24b0
  do_exit+0x7bd/0x24b0
  ? __pfx_migrate_enable+0x10/0x10
  ? __pfx_do_exit+0x10/0x10
  ? __pfx_read_tsc+0x10/0x10
  ? ktime_get+0x64/0x140
  ? _raw_spin_lock_irq+0x86/0xe0
  do_group_exit+0xb0/0x220
  get_signal+0x17ba/0x1b50
  ? vfs_read+0x179/0xa40
  ? timerlat_fd_read+0x30b/0x9d0
  ? __pfx_get_signal+0x10/0x10
  ? __pfx_timerlat_fd_read+0x10/0x10
  arch_do_signal_or_restart+0x8c/0x570
  ? __pfx_arch_do_signal_or_restart+0x10/0x10
  ? vfs_read+0x179/0xa40
  ? ksys_read+0xfe/0x1d0
  ? __pfx_ksys_read+0x10/0x10
  syscall_exit_to_user_mode+0xbc/0x130
  do_syscall_64+0x74/0x110
  ? __pfx___rseq_handle_notify_resume+0x10/0x10
  ? __pfx_ksys_read+0x10/0x10
  ? fpregs_restore_userregs+0xdb/0x1e0
  ? fpregs_restore_userregs+0xdb/0x1e0
  ? syscall_exit_to_user_mode+0x116/0x130
  ? do_syscall_64+0x74/0x110
  ? do_syscall_64+0x74/0x110
  ? do_syscall_64+0x74/0x110
  entry_SYSCALL_64_after_hwframe+0x71/0x79
 RIP: 0033:0x7ff0070eca9c
 Code: Unable to access opcode bytes at 0x7ff0070eca72.
 RSP: 002b:00007ff006dff8c0 EFLAGS: 00000246 ORIG_RAX: 0000000000000000
 RAX: 0000000000000000 RBX: 0000000000000005 RCX: 00007ff0070eca9c
 RDX: 0000000000000400 RSI: 00007ff006dff9a0 RDI: 0000000000000003
 RBP: 00007ff006dffde0 R08: 0000000000000000 R09: 00007ff000000ba0
 R10: 00007ff007004b08 R11: 0000000000000246 R12: 0000000000000003
 R13: 00007ff006dff9a0 R14: 0000000000000007 R15: 0000000000000008
  </TASK>
 Modules linked in: snd_hda_intel snd_intel_dspcfg snd_intel_sdw_acpi snd_hda_codec snd_hwdep snd_hda_core
 ---[ end trace 0000000000000000 ]---

This is because it would mistakenly call kthread_stop() on a user space
thread making it "exit" before it actually exits.

Since kthreads are created based on global behavior, use a cpumask to know
when kthreads are running and that they need to be shutdown before
proceeding to do new work.

Link: https://lore.kernel.org/all/[email protected]/
This was debugged by using the persistent ring buffer:

Link: https://lore.kernel.org/all/[email protected]/
Note, locking was originally used to fix this, but that proved to cause too
many deadlocks to work around:

  https://lore.kernel.org/linux-trace-kernel/20240823102816.5e55753b@gandalf.local.home/

Cc: [email protected]
Cc: Masami Hiramatsu <[email protected]>
Cc: Mathieu Desnoyers <[email protected]>
Cc: "Luis Claudio R. Goncalves" <[email protected]>
Link: https://lore.kernel.org/[email protected]
Fixes: e88ed227f639e ("tracing/timerlat: Add user-space interface")
Reported-by: Tomas Glozar <[email protected]>
Signed-off-by: Steven Rostedt (Google) <[email protected]>
6 months agoeventfs: Use list_del_rcu() for SRCU protected list variable
Steven Rostedt [Wed, 4 Sep 2024 17:16:05 +0000 (13:16 -0400)]
eventfs: Use list_del_rcu() for SRCU protected list variable

Chi Zhiling reported:

  We found a null pointer accessing in tracefs[1], the reason is that the
  variable 'ei_child' is set to LIST_POISON1, that means the list was
  removed in eventfs_remove_rec. so when access the ei_child->is_freed, the
  panic triggered.

  by the way, the following script can reproduce this panic

  loop1 (){
      while true
      do
          echo "p:kp submit_bio" > /sys/kernel/debug/tracing/kprobe_events
          echo "" > /sys/kernel/debug/tracing/kprobe_events
      done
  }
  loop2 (){
      while true
      do
          tree /sys/kernel/debug/tracing/events/kprobes/
      done
  }
  loop1 &
  loop2

  [1]:
  [ 1147.959632][T17331] Unable to handle kernel paging request at virtual address dead000000000150
  [ 1147.968239][T17331] Mem abort info:
  [ 1147.971739][T17331]   ESR = 0x0000000096000004
  [ 1147.976172][T17331]   EC = 0x25: DABT (current EL), IL = 32 bits
  [ 1147.982171][T17331]   SET = 0, FnV = 0
  [ 1147.985906][T17331]   EA = 0, S1PTW = 0
  [ 1147.989734][T17331]   FSC = 0x04: level 0 translation fault
  [ 1147.995292][T17331] Data abort info:
  [ 1147.998858][T17331]   ISV = 0, ISS = 0x00000004, ISS2 = 0x00000000
  [ 1148.005023][T17331]   CM = 0, WnR = 0, TnD = 0, TagAccess = 0
  [ 1148.010759][T17331]   GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0
  [ 1148.016752][T17331] [dead000000000150] address between user and kernel address ranges
  [ 1148.024571][T17331] Internal error: Oops: 0000000096000004 [#1] SMP
  [ 1148.030825][T17331] Modules linked in: team_mode_loadbalance team nlmon act_gact cls_flower sch_ingress bonding tls macvlan dummy ib_core bridge stp llc veth amdgpu amdxcp mfd_core gpu_sched drm_exec drm_buddy radeon crct10dif_ce video drm_suballoc_helper ghash_ce drm_ttm_helper sha2_ce ttm sha256_arm64 i2c_algo_bit sha1_ce sbsa_gwdt cp210x drm_display_helper cec sr_mod cdrom drm_kms_helper binfmt_misc sg loop fuse drm dm_mod nfnetlink ip_tables autofs4 [last unloaded: tls]
  [ 1148.072808][T17331] CPU: 3 PID: 17331 Comm: ls Tainted: G        W         ------- ----  6.6.43 #2
  [ 1148.081751][T17331] Source Version: 21b3b386e948bedd29369af66f3e98ab01b1c650
  [ 1148.088783][T17331] Hardware name: Greatwall GW-001M1A-FTF/GW-001M1A-FTF, BIOS KunLun BIOS V4.0 07/16/2020
  [ 1148.098419][T17331] pstate: 20000005 (nzCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
  [ 1148.106060][T17331] pc : eventfs_iterate+0x2c0/0x398
  [ 1148.111017][T17331] lr : eventfs_iterate+0x2fc/0x398
  [ 1148.115969][T17331] sp : ffff80008d56bbd0
  [ 1148.119964][T17331] x29: ffff80008d56bbf0 x28: ffff001ff5be2600 x27: 0000000000000000
  [ 1148.127781][T17331] x26: ffff001ff52ca4e0 x25: 0000000000009977 x24: dead000000000100
  [ 1148.135598][T17331] x23: 0000000000000000 x22: 000000000000000b x21: ffff800082645f10
  [ 1148.143415][T17331] x20: ffff001fddf87c70 x19: ffff80008d56bc90 x18: 0000000000000000
  [ 1148.151231][T17331] x17: 0000000000000000 x16: 0000000000000000 x15: ffff001ff52ca4e0
  [ 1148.159048][T17331] x14: 0000000000000000 x13: 0000000000000000 x12: 0000000000000000
  [ 1148.166864][T17331] x11: 0000000000000000 x10: 0000000000000000 x9 : ffff8000804391d0
  [ 1148.174680][T17331] x8 : 0000000180000000 x7 : 0000000000000018 x6 : 0000aaab04b92862
  [ 1148.182498][T17331] x5 : 0000aaab04b92862 x4 : 0000000080000000 x3 : 0000000000000068
  [ 1148.190314][T17331] x2 : 000000000000000f x1 : 0000000000007ea8 x0 : 0000000000000001
  [ 1148.198131][T17331] Call trace:
  [ 1148.201259][T17331]  eventfs_iterate+0x2c0/0x398
  [ 1148.205864][T17331]  iterate_dir+0x98/0x188
  [ 1148.210036][T17331]  __arm64_sys_getdents64+0x78/0x160
  [ 1148.215161][T17331]  invoke_syscall+0x78/0x108
  [ 1148.219593][T17331]  el0_svc_common.constprop.0+0x48/0xf0
  [ 1148.224977][T17331]  do_el0_svc+0x24/0x38
  [ 1148.228974][T17331]  el0_svc+0x40/0x168
  [ 1148.232798][T17331]  el0t_64_sync_handler+0x120/0x130
  [ 1148.237836][T17331]  el0t_64_sync+0x1a4/0x1a8
  [ 1148.242182][T17331] Code: 54ffff6c f9400676 910006d6 f9000676 (b9405300)
  [ 1148.248955][T17331] ---[ end trace 0000000000000000 ]---

The issue is that list_del() is used on an SRCU protected list variable
before the synchronization occurs. This can poison the list pointers while
there is a reader iterating the list.

This is simply fixed by using list_del_rcu() that is specifically made for
this purpose.

Link: https://lore.kernel.org/linux-trace-kernel/[email protected]/
Cc: [email protected]
Cc: Masami Hiramatsu <[email protected]>
Cc: Mathieu Desnoyers <[email protected]>
Link: https://lore.kernel.org/[email protected]
Fixes: 43aa6f97c2d03 ("eventfs: Get rid of dentry pointers without refcounts")
Reported-by: Chi Zhiling <[email protected]>
Tested-by: Chi Zhiling <[email protected]>
Signed-off-by: Steven Rostedt (Google) <[email protected]>
6 months agotracing: Avoid possible softlockup in tracing_iter_reset()
Zheng Yejian [Tue, 27 Aug 2024 12:46:54 +0000 (20:46 +0800)]
tracing: Avoid possible softlockup in tracing_iter_reset()

In __tracing_open(), when max latency tracers took place on the cpu,
the time start of its buffer would be updated, then event entries with
timestamps being earlier than start of the buffer would be skipped
(see tracing_iter_reset()).

Softlockup will occur if the kernel is non-preemptible and too many
entries were skipped in the loop that reset every cpu buffer, so add
cond_resched() to avoid it.

Cc: [email protected]
Fixes: 2f26ebd549b9a ("tracing: use timestamp to determine start of latency traces")
Link: https://lore.kernel.org/[email protected]
Suggested-by: Steven Rostedt <[email protected]>
Signed-off-by: Zheng Yejian <[email protected]>
Signed-off-by: Steven Rostedt (Google) <[email protected]>
6 months agoMerge tag 'bcachefs-2024-09-04' of git://evilpiepirate.org/bcachefs
Linus Torvalds [Wed, 4 Sep 2024 20:54:47 +0000 (13:54 -0700)]
Merge tag 'bcachefs-2024-09-04' of git://evilpiepirate.org/bcachefs

Pull bcachefs fixes from Kent Overstreet:

 - Fix a typo in the rebalance accounting changes

 - BCH_SB_MEMBER_INVALID: small on disk format feature which will be
   needed for full erasure coding support; this is only the minimum so
   that 6.11 can handle future versions without barfing.

* tag 'bcachefs-2024-09-04' of git://evilpiepirate.org/bcachefs:
  bcachefs: BCH_SB_MEMBER_INVALID
  bcachefs: fix rebalance accounting

6 months agoMerge tag 'perf-tools-fixes-for-v6.11-2024-09-04' of git://git.kernel.org/pub/scm...
Linus Torvalds [Wed, 4 Sep 2024 19:10:19 +0000 (12:10 -0700)]
Merge tag 'perf-tools-fixes-for-v6.11-2024-09-04' of git://git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools

Pull perf tools fixes from Namhyung Kim:
 "A number of small fixes for the late cycle:

   - Two more build fixes on 32-bit archs

   - Fixed a segfault during perf test

   - Fixed spinlock/rwlock accounting bug in perf lock contention"

* tag 'perf-tools-fixes-for-v6.11-2024-09-04' of git://git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools:
  perf daemon: Fix the build on more 32-bit architectures
  perf python: include "util/sample.h"
  perf lock contention: Fix spinlock and rwlock accounting
  perf test pmu: Set uninitialized PMU alias to null

6 months agoMerge tag 'hwmon-for-v6.11-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 4 Sep 2024 19:07:31 +0000 (12:07 -0700)]
Merge tag 'hwmon-for-v6.11-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging

Pull hwmon fixes from Guenter Roeck:

 - hp-wmi-sensors: Check if WMI event data exists before accessing it

 - ltc2991: fix register bits defines

* tag 'hwmon-for-v6.11-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  hwmon: (hp-wmi-sensors) Check if WMI event data exists
  hwmon: ltc2991: fix register bits defines

6 months agoMerge tag 'for-6.11-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave...
Linus Torvalds [Wed, 4 Sep 2024 18:53:47 +0000 (11:53 -0700)]
Merge tag 'for-6.11-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux

Pull btrfs fixes from David Sterba:

 - followup fix for direct io and fsync under some conditions, reported
   by QEMU users

 - fix a potential leak when disabling quotas while some extent tracking
   work can still happen

 - in zoned mode handle unexpected change of zone write pointer in
   RAID1-like block groups, turn the zones to read-only

* tag 'for-6.11-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
  btrfs: fix race between direct IO write and fsync when using same fd
  btrfs: zoned: handle broken write pointer on zones
  btrfs: qgroup: don't use extent changeset when not needed

6 months agoMerge tag 'v6.11-rc6-server-fixes' of git://git.samba.org/ksmbd
Linus Torvalds [Wed, 4 Sep 2024 16:41:51 +0000 (09:41 -0700)]
Merge tag 'v6.11-rc6-server-fixes' of git://git.samba.org/ksmbd

Pull smb server fixes from Steve French:

 - Fix crash in session setup

 - Fix locking bug

 - Improve access bounds checking

* tag 'v6.11-rc6-server-fixes' of git://git.samba.org/ksmbd:
  ksmbd: Unlock on in ksmbd_tcp_set_interfaces()
  ksmbd: unset the binding mark of a reused connection
  smb: Annotate struct xattr_smb_acl with __counted_by()

6 months agoMerge tag 'vfs-6.11-rc7.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs
Linus Torvalds [Wed, 4 Sep 2024 16:33:57 +0000 (09:33 -0700)]
Merge tag 'vfs-6.11-rc7.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs

Pull vfs fixes from Christian Brauner:
 "Two netfs fixes for this merge window:

   - Ensure that fscache_cookie_lru_time is deleted when the fscache
     module is removed to prevent UAF

   - Fix filemap_invalidate_inode() to use invalidate_inode_pages2_range()

     Before it used truncate_inode_pages_partial() which causes
     copy_file_range() to fail on cifs"

* tag 'vfs-6.11-rc7.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs:
  fscache: delete fscache_cookie_lru_timer when fscache exits to avoid UAF
  mm: Fix filemap_invalidate_inode() to use invalidate_inode_pages2_range()

6 months agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rmk/linux
Linus Torvalds [Wed, 4 Sep 2024 16:17:33 +0000 (09:17 -0700)]
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rmk/linux

Pull ARM fix from Russell King:

 - Fix a build issue with older binutils with LD dead code elimination
   disabled

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rmk/linux:
  ARM: 9414/1: Fix build issue with LD_DEAD_CODE_DATA_ELIMINATION

6 months agoMerge tag 'parisc-for-6.11-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 4 Sep 2024 16:11:57 +0000 (09:11 -0700)]
Merge tag 'parisc-for-6.11-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux

Pull parisc architecture fix from Helge Deller:

 - Fix boot issue where boot memory is marked read-only too early

* tag 'parisc-for-6.11-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: Delay write-protection until mark_rodata_ro() call

6 months agoMerge tag 'mm-hotfixes-stable-2024-09-03-20-19' of git://git.kernel.org/pub/scm/linux...
Linus Torvalds [Wed, 4 Sep 2024 15:37:33 +0000 (08:37 -0700)]
Merge tag 'mm-hotfixes-stable-2024-09-03-20-19' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm

Pull misc fixes from Andrew Morton:
 "17 hotfixes, 15 of which are cc:stable.

  Mostly MM, no identifiable theme.  And a few nilfs2 fixups"

* tag 'mm-hotfixes-stable-2024-09-03-20-19' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm:
  alloc_tag: fix allocation tag reporting when CONFIG_MODULES=n
  mm: vmalloc: optimize vmap_lazy_nr arithmetic when purging each vmap_area
  mailmap: update entry for Jan Kuliga
  codetag: debug: mark codetags for poisoned page as empty
  mm/memcontrol: respect zswap.writeback setting from parent cg too
  scripts: fix gfp-translate after ___GFP_*_BITS conversion to an enum
  Revert "mm: skip CMA pages when they are not available"
  maple_tree: remove rcu_read_lock() from mt_validate()
  kexec_file: fix elfcorehdr digest exclusion when CONFIG_CRASH_HOTPLUG=y
  mm/slub: add check for s->flags in the alloc_tagging_slab_free_hook
  nilfs2: fix state management in error path of log writing function
  nilfs2: fix missing cleanup on rollforward recovery error
  nilfs2: protect references to superblock parameters exposed in sysfs
  userfaultfd: don't BUG_ON() if khugepaged yanks our page table
  userfaultfd: fix checks for huge PMDs
  mm: vmalloc: ensure vmap_block is initialised before adding to queue
  selftests: mm: fix build errors on armhf

6 months agoARM: 9414/1: Fix build issue with LD_DEAD_CODE_DATA_ELIMINATION
Yuntao Liu [Wed, 21 Aug 2024 06:34:41 +0000 (07:34 +0100)]
ARM: 9414/1: Fix build issue with LD_DEAD_CODE_DATA_ELIMINATION

There is a build issue with LD segmentation fault, while
CONFIG_LD_DEAD_CODE_DATA_ELIMINATION is not enabled, as bellow.

scripts/link-vmlinux.sh: line 49:  3796 Segmentation fault
 (core dumped) ${ld} ${ldflags} -o ${output} ${wl}--whole-archive
 ${objs} ${wl}--no-whole-archive ${wl}--start-group
 ${libs} ${wl}--end-group ${kallsymso} ${btf_vmlinux_bin_o} ${ldlibs}

The error occurs in older versions of the GNU ld with version earlier
than 2.36. It makes most sense to have a minimum LD version as
a dependency for HAVE_LD_DEAD_CODE_DATA_ELIMINATION and eliminate
the impact of ".reloc  .text, R_ARM_NONE, ." when
CONFIG_LD_DEAD_CODE_DATA_ELIMINATION is not enabled.

Fixes: ed0f94102251 ("ARM: 9404/1: arm32: enable HAVE_LD_DEAD_CODE_DATA_ELIMINATION")
Reported-by: Harith George <[email protected]>
Tested-by: Harith George <[email protected]>
Suggested-by: Arnd Bergmann <[email protected]>
Acked-by: Arnd Bergmann <[email protected]>
Signed-off-by: Yuntao Liu <[email protected]>
Link: https://lore.kernel.org/all/[email protected]/
Signed-off-by: Russell King (Oracle) <[email protected]>
6 months agobcachefs: BCH_SB_MEMBER_INVALID
Kent Overstreet [Sun, 1 Sep 2024 22:09:18 +0000 (18:09 -0400)]
bcachefs: BCH_SB_MEMBER_INVALID

Create a sentinal value for "invalid device".

This is needed for removing devices that have stripes on them (force
removing, without evacuating); we need a sentinal value for the stripe
pointers to the device being removed.

Signed-off-by: Kent Overstreet <[email protected]>
6 months agoMAINTAINERS: update Andreas Hindborg's email address
Andreas Hindborg [Tue, 3 Sep 2024 20:09:48 +0000 (22:09 +0200)]
MAINTAINERS: update Andreas Hindborg's email address

Move away from corporate infrastructure for upstream work. Also update
mailmap.

Signed-off-by: Andreas Hindborg <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
[ Reworded title slightly. - Miguel ]
Signed-off-by: Miguel Ojeda <[email protected]>
6 months agoMerge tag 'fuse-fixes-6.11-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 3 Sep 2024 19:32:00 +0000 (12:32 -0700)]
Merge tag 'fuse-fixes-6.11-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse

Pull fuse fixes from Miklos Szeredi:

 - Fix EIO if splice and page stealing are enabled on the fuse device

 - Disable problematic combination of passthrough and writeback-cache

 - Other bug fixes found by code review

* tag 'fuse-fixes-6.11-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse:
  fuse: disable the combination of passthrough and writeback cache
  fuse: update stats for pages in dropped aux writeback list
  fuse: clear PG_uptodate when using a stolen page
  fuse: fix memory leak in fuse_create_open
  fuse: check aborted connection before adding requests to pending list for resending
  fuse: use unsigned type for getxattr/listxattr size truncation

6 months agobtrfs: fix race between direct IO write and fsync when using same fd
Filipe Manana [Thu, 29 Aug 2024 17:25:49 +0000 (18:25 +0100)]
btrfs: fix race between direct IO write and fsync when using same fd

If we have 2 threads that are using the same file descriptor and one of
them is doing direct IO writes while the other is doing fsync, we have a
race where we can end up either:

1) Attempt a fsync without holding the inode's lock, triggering an
   assertion failures when assertions are enabled;

2) Do an invalid memory access from the fsync task because the file private
   points to memory allocated on stack by the direct IO task and it may be
   used by the fsync task after the stack was destroyed.

The race happens like this:

1) A user space program opens a file descriptor with O_DIRECT;

2) The program spawns 2 threads using libpthread for example;

3) One of the threads uses the file descriptor to do direct IO writes,
   while the other calls fsync using the same file descriptor.

4) Call task A the thread doing direct IO writes and task B the thread
   doing fsyncs;

5) Task A does a direct IO write, and at btrfs_direct_write() sets the
   file's private to an on stack allocated private with the member
   'fsync_skip_inode_lock' set to true;

6) Task B enters btrfs_sync_file() and sees that there's a private
   structure associated to the file which has 'fsync_skip_inode_lock' set
   to true, so it skips locking the inode's VFS lock;

7) Task A completes the direct IO write, and resets the file's private to
   NULL since it had no prior private and our private was stack allocated.
   Then it unlocks the inode's VFS lock;

8) Task B enters btrfs_get_ordered_extents_for_logging(), then the
   assertion that checks the inode's VFS lock is held fails, since task B
   never locked it and task A has already unlocked it.

The stack trace produced is the following:

   assertion failed: inode_is_locked(&inode->vfs_inode), in fs/btrfs/ordered-data.c:983
   ------------[ cut here ]------------
   kernel BUG at fs/btrfs/ordered-data.c:983!
   Oops: invalid opcode: 0000 [#1] PREEMPT SMP PTI
   CPU: 9 PID: 5072 Comm: worker Tainted: G     U     OE      6.10.5-1-default #1 openSUSE Tumbleweed 69f48d427608e1c09e60ea24c6c55e2ca1b049e8
   Hardware name: Acer Predator PH315-52/Covini_CFS, BIOS V1.12 07/28/2020
   RIP: 0010:btrfs_get_ordered_extents_for_logging.cold+0x1f/0x42 [btrfs]
   Code: 50 d6 86 c0 e8 (...)
   RSP: 0018:ffff9e4a03dcfc78 EFLAGS: 00010246
   RAX: 0000000000000054 RBX: ffff9078a9868e98 RCX: 0000000000000000
   RDX: 0000000000000000 RSI: ffff907dce4a7800 RDI: ffff907dce4a7800
   RBP: ffff907805518800 R08: 0000000000000000 R09: ffff9e4a03dcfb38
   R10: ffff9e4a03dcfb30 R11: 0000000000000003 R12: ffff907684ae7800
   R13: 0000000000000001 R14: ffff90774646b600 R15: 0000000000000000
   FS:  00007f04b96006c0(0000) GS:ffff907dce480000(0000) knlGS:0000000000000000
   CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
   CR2: 00007f32acbfc000 CR3: 00000001fd4fa005 CR4: 00000000003726f0
   Call Trace:
    <TASK>
    ? __die_body.cold+0x14/0x24
    ? die+0x2e/0x50
    ? do_trap+0xca/0x110
    ? do_error_trap+0x6a/0x90
    ? btrfs_get_ordered_extents_for_logging.cold+0x1f/0x42 [btrfs bb26272d49b4cdc847cf3f7faadd459b62caee9a]
    ? exc_invalid_op+0x50/0x70
    ? btrfs_get_ordered_extents_for_logging.cold+0x1f/0x42 [btrfs bb26272d49b4cdc847cf3f7faadd459b62caee9a]
    ? asm_exc_invalid_op+0x1a/0x20
    ? btrfs_get_ordered_extents_for_logging.cold+0x1f/0x42 [btrfs bb26272d49b4cdc847cf3f7faadd459b62caee9a]
    ? btrfs_get_ordered_extents_for_logging.cold+0x1f/0x42 [btrfs bb26272d49b4cdc847cf3f7faadd459b62caee9a]
    btrfs_sync_file+0x21a/0x4d0 [btrfs bb26272d49b4cdc847cf3f7faadd459b62caee9a]
    ? __seccomp_filter+0x31d/0x4f0
    __x64_sys_fdatasync+0x4f/0x90
    do_syscall_64+0x82/0x160
    ? do_futex+0xcb/0x190
    ? __x64_sys_futex+0x10e/0x1d0
    ? switch_fpu_return+0x4f/0xd0
    ? syscall_exit_to_user_mode+0x72/0x220
    ? do_syscall_64+0x8e/0x160
    ? syscall_exit_to_user_mode+0x72/0x220
    ? do_syscall_64+0x8e/0x160
    ? syscall_exit_to_user_mode+0x72/0x220
    ? do_syscall_64+0x8e/0x160
    ? syscall_exit_to_user_mode+0x72/0x220
    ? do_syscall_64+0x8e/0x160
    entry_SYSCALL_64_after_hwframe+0x76/0x7e

Another problem here is if task B grabs the private pointer and then uses
it after task A has finished, since the private was allocated in the stack
of task A, it results in some invalid memory access with a hard to predict
result.

This issue, triggering the assertion, was observed with QEMU workloads by
two users in the Link tags below.

Fix this by not relying on a file's private to pass information to fsync
that it should skip locking the inode and instead pass this information
through a special value stored in current->journal_info. This is safe
because in the relevant section of the direct IO write path we are not
holding a transaction handle, so current->journal_info is NULL.

The following C program triggers the issue:

   $ cat repro.c
   /* Get the O_DIRECT definition. */
   #ifndef _GNU_SOURCE
   #define _GNU_SOURCE
   #endif

   #include <stdio.h>
   #include <stdlib.h>
   #include <unistd.h>
   #include <stdint.h>
   #include <fcntl.h>
   #include <errno.h>
   #include <string.h>
   #include <pthread.h>

   static int fd;

   static ssize_t do_write(int fd, const void *buf, size_t count, off_t offset)
   {
       while (count > 0) {
           ssize_t ret;

           ret = pwrite(fd, buf, count, offset);
           if (ret < 0) {
               if (errno == EINTR)
                   continue;
               return ret;
           }
           count -= ret;
           buf += ret;
       }
       return 0;
   }

   static void *fsync_loop(void *arg)
   {
       while (1) {
           int ret;

           ret = fsync(fd);
           if (ret != 0) {
               perror("Fsync failed");
               exit(6);
           }
       }
   }

   int main(int argc, char *argv[])
   {
       long pagesize;
       void *write_buf;
       pthread_t fsyncer;
       int ret;

       if (argc != 2) {
           fprintf(stderr, "Use: %s <file path>\n", argv[0]);
           return 1;
       }

       fd = open(argv[1], O_WRONLY | O_CREAT | O_TRUNC | O_DIRECT, 0666);
       if (fd == -1) {
           perror("Failed to open/create file");
           return 1;
       }

       pagesize = sysconf(_SC_PAGE_SIZE);
       if (pagesize == -1) {
           perror("Failed to get page size");
           return 2;
       }

       ret = posix_memalign(&write_buf, pagesize, pagesize);
       if (ret) {
           perror("Failed to allocate buffer");
           return 3;
       }

       ret = pthread_create(&fsyncer, NULL, fsync_loop, NULL);
       if (ret != 0) {
           fprintf(stderr, "Failed to create writer thread: %d\n", ret);
           return 4;
       }

       while (1) {
           ret = do_write(fd, write_buf, pagesize, 0);
           if (ret != 0) {
               perror("Write failed");
               exit(5);
           }
       }

       return 0;
   }

   $ mkfs.btrfs -f /dev/sdi
   $ mount /dev/sdi /mnt/sdi
   $ timeout 10 ./repro /mnt/sdi/foo

Usually the race is triggered within less than 1 second. A test case for
fstests will follow soon.

Reported-by: Paulo Dias <[email protected]>
Link: https://bugzilla.kernel.org/show_bug.cgi?id=219187
Reported-by: Andreas Jahn <[email protected]>
Link: https://bugzilla.kernel.org/show_bug.cgi?id=219199
Reported-by: [email protected]
Link: https://lore.kernel.org/linux-btrfs/[email protected]/
Fixes: 939b656bc8ab ("btrfs: fix corruption after buffer fault in during direct IO append write")
CC: [email protected] # 5.15+
Reviewed-by: Josef Bacik <[email protected]>
Signed-off-by: Filipe Manana <[email protected]>
Reviewed-by: David Sterba <[email protected]>
Signed-off-by: David Sterba <[email protected]>
6 months agoplatform/x86/amd: pmf: Make ASUS GA403 quirk generic
Luke D. Jones [Sat, 31 Aug 2024 00:39:05 +0000 (12:39 +1200)]
platform/x86/amd: pmf: Make ASUS GA403 quirk generic

The original quirk should match to GA403U so that the full
range of GA403U models can benefit.

Signed-off-by: Luke D. Jones <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Reviewed-by: Ilpo Järvinen <[email protected]>
Signed-off-by: Ilpo Järvinen <[email protected]>
6 months agoparisc: Delay write-protection until mark_rodata_ro() call
Helge Deller [Sat, 31 Aug 2024 12:02:06 +0000 (14:02 +0200)]
parisc: Delay write-protection until mark_rodata_ro() call

Do not write-protect the kernel read-only and __ro_after_init sections
earlier than before mark_rodata_ro() is called.  This fixes a boot issue on
parisc which is triggered by commit 91a1d97ef482 ("jump_label,module: Don't
alloc static_key_mod for __ro_after_init keys"). That commit may modify
static key contents in the __ro_after_init section at bootup, so this
section needs to be writable at least until mark_rodata_ro() is called.

Signed-off-by: Helge Deller <[email protected]>
Reported-by: matoro <[email protected]>
Reported-by: Christoph Biedl <[email protected]>
Tested-by: Christoph Biedl <[email protected]>
Link: https://lore.kernel.org/linux-parisc/[email protected]/#r
Fixes: 91a1d97ef482 ("jump_label,module: Don't alloc static_key_mod for __ro_after_init keys")
Cc: [email protected] # v6.10+
6 months agobtrfs: zoned: handle broken write pointer on zones
Naohiro Aota [Fri, 30 Aug 2024 16:32:49 +0000 (01:32 +0900)]
btrfs: zoned: handle broken write pointer on zones

Btrfs rejects to mount a FS if it finds a block group with a broken write
pointer (e.g, unequal write pointers on two zones of RAID1 block group).
Since such case can happen easily with a power-loss or crash of a system,
we need to handle the case more gently.

Handle such block group by making it unallocatable, so that there will be
no writes into it. That can be done by setting the allocation pointer at
the end of allocating region (= block_group->zone_capacity). Then, existing
code handle zone_unusable properly.

Having proper zone_capacity is necessary for the change. So, set it as fast
as possible.

We cannot handle RAID0 and RAID10 case like this. But, they are anyway
unable to read because of a missing stripe.

Fixes: 265f7237dd25 ("btrfs: zoned: allow DUP on meta-data block groups")
Fixes: 568220fa9657 ("btrfs: zoned: support RAID0/1/10 on top of raid stripe tree")
CC: [email protected] # 6.1+
Reported-by: HAN Yuwei <[email protected]>
Cc: Xuefer <[email protected]>
Signed-off-by: Naohiro Aota <[email protected]>
Signed-off-by: David Sterba <[email protected]>
6 months agoperf daemon: Fix the build on more 32-bit architectures
Arnaldo Carvalho de Melo [Tue, 20 Aug 2024 00:43:01 +0000 (21:43 -0300)]
perf daemon: Fix the build on more 32-bit architectures

FYI: I'm carrying this on perf-tools-next.

The previous attempt fixed the build on debian:experimental-x-mipsel,
but when building on a larger set of containers I noticed it broke the
build on some other 32-bit architectures such as:

  42     7.87 ubuntu:18.04-x-arm            : FAIL gcc version 7.5.0 (Ubuntu/Linaro 7.5.0-3ubuntu1~18.04)
    builtin-daemon.c: In function 'cmd_session_list':
    builtin-daemon.c:692:16: error: format '%llu' expects argument of type 'long long unsigned int', but argument 4 has type 'long int' [-Werror=format=]
       fprintf(out, "%c%" PRIu64,
                    ^~~~~
    builtin-daemon.c:694:13:
        csv_sep, (curr - daemon->start) / 60);
                 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
    In file included from builtin-daemon.c:3:0:
    /usr/arm-linux-gnueabihf/include/inttypes.h:105:34: note: format string is defined here
     # define PRIu64  __PRI64_PREFIX "u"

So lets cast that time_t (32-bit/64-bit) to uint64_t to make sure it
builds everywhere.

Fixes: 4bbe6002931954bb ("perf daemon: Fix the build on 32-bit architectures")
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
Link: https://lore.kernel.org/r/ZsPmldtJ0D9Cua9_@x1
Signed-off-by: Namhyung Kim <[email protected]>
6 months agoperf python: include "util/sample.h"
Xu Yang [Mon, 19 Aug 2024 02:34:03 +0000 (10:34 +0800)]
perf python: include "util/sample.h"

The 32-bit arm build system will complain:

tools/perf/util/python.c:75:28: error: field ‘sample’ has incomplete type
   75 |         struct perf_sample sample;

However, arm64 build system doesn't complain this.

The root cause is arm64 define "HAVE_KVM_STAT_SUPPORT := 1" in
tools/perf/arch/arm64/Makefile, but arm arch doesn't define this.
This will lead to kvm-stat.h include other header files on arm64 build
system, especially "util/sample.h" for util/python.c.

This will try to directly include "util/sample.h" for "util/python.c" to
avoid such build issue on arm platform.

Signed-off-by: Xu Yang <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Namhyung Kim <[email protected]>
6 months agoperf lock contention: Fix spinlock and rwlock accounting
Namhyung Kim [Wed, 28 Aug 2024 05:29:53 +0000 (22:29 -0700)]
perf lock contention: Fix spinlock and rwlock accounting

The spinlock and rwlock use a single-element per-cpu array to track
current locks due to performance reason.  But this means the key is
always available and it cannot simply account lock stats in the array
because some of them are invalid.

In fact, the contention_end() program in the BPF invalidates the entry
by setting the 'lock' value to 0 instead of deleting the entry for the
hashmap.  So it should skip entries with the lock value of 0 in the
account_end_timestamp().

Otherwise, it'd have spurious high contention on an idle machine:

  $ sudo perf lock con -ab -Y spinlock sleep 3
   contended   total wait     max wait     avg wait         type   caller

           8      4.72 s       1.84 s     590.46 ms     spinlock   rcu_core+0xc7
           8      1.87 s       1.87 s     233.48 ms     spinlock   process_one_work+0x1b5
           2      1.87 s       1.87 s     933.92 ms     spinlock   worker_thread+0x1a2
           3      1.81 s       1.81 s     603.93 ms     spinlock   tmigr_update_events+0x13c
           2      1.72 s       1.72 s     861.98 ms     spinlock   tick_do_update_jiffies64+0x25
           6     42.48 us     13.02 us      7.08 us     spinlock   futex_q_lock+0x2a
           1     13.03 us     13.03 us     13.03 us     spinlock   futex_wake+0xce
           1     11.61 us     11.61 us     11.61 us     spinlock   rcu_core+0xc7

I don't believe it has contention on a spinlock longer than 1 second.
After this change, it only reports some small contentions.

  $ sudo perf lock con -ab -Y spinlock sleep 3
   contended   total wait     max wait     avg wait         type   caller

           4    133.51 us     43.29 us     33.38 us     spinlock   tick_do_update_jiffies64+0x25
           4     69.06 us     31.82 us     17.27 us     spinlock   process_one_work+0x1b5
           2     50.66 us     25.77 us     25.33 us     spinlock   rcu_core+0xc7
           1     28.45 us     28.45 us     28.45 us     spinlock   rcu_core+0xc7
           1     24.77 us     24.77 us     24.77 us     spinlock   tmigr_update_events+0x13c
           1     23.34 us     23.34 us     23.34 us     spinlock   raw_spin_rq_lock_nested+0x15

Fixes: b5711042a1c8 ("perf lock contention: Use per-cpu array map for spinlocks")
Reported-by: Xi Wang <[email protected]>
Cc: Song Liu <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Namhyung Kim <[email protected]>
6 months agoperf test pmu: Set uninitialized PMU alias to null
Veronika Molnarova [Thu, 8 Aug 2024 10:37:49 +0000 (12:37 +0200)]
perf test pmu: Set uninitialized PMU alias to null

Commit 3e0bf9fde2984469 ("perf pmu: Restore full PMU name wildcard
support") adds a test case "PMU cmdline match" that covers PMU name
wildcard support provided by function perf_pmu__match(). The test works
with a wide range of supported combinations of PMU name matching but
omits the case that if the perf_pmu__match() cannot match the PMU name
to the wildcard, it tries to match its alias. However, this variable is
not set up, causing the test case to fail when run with subprocesses or
to segfault if run as a single process.

  ./perf test -vv 9
    9: Sysfs PMU tests                                                 :
    9.1: Parsing with PMU format directory                             : Ok
    9.2: Parsing with PMU event                                        : Ok
    9.3: PMU event names                                               : Ok
    9.4: PMU name combining                                            : Ok
    9.5: PMU name comparison                                           : Ok
    9.6: PMU cmdline match                                             : FAILED!

  ./perf test -F 9
    9.1: Parsing with PMU format directory                             : Ok
    9.2: Parsing with PMU event                                        : Ok
    9.3: PMU event names                                               : Ok
    9.4: PMU name combining                                            : Ok
    9.5: PMU name comparison                                           : Ok
  Segmentation fault (core dumped)

Initialize the PMU alias to null for all tests of perf_pmu__match()
as this functionality is not being tested and the alias matching works
exactly the same as the matching of the PMU name.

  ./perf test -F 9
    9.1: Parsing with PMU format directory                             : Ok
    9.2: Parsing with PMU event                                        : Ok
    9.3: PMU event names                                               : Ok
    9.4: PMU name combining                                            : Ok
    9.5: PMU name comparison                                           : Ok
    9.6: PMU cmdline match                                             : Ok

Fixes: 3e0bf9fde2984469 ("perf pmu: Restore full PMU name wildcard support")
Signed-off-by: Veronika Molnarova <[email protected]>
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Namhyung Kim <[email protected]>
6 months agobtrfs: qgroup: don't use extent changeset when not needed
Fedor Pchelkin [Wed, 28 Aug 2024 16:14:11 +0000 (19:14 +0300)]
btrfs: qgroup: don't use extent changeset when not needed

The local extent changeset is passed to clear_record_extent_bits() where
it may have some additional memory dynamically allocated for ulist. When
qgroup is disabled, the memory is leaked because in this case the
changeset is not released upon __btrfs_qgroup_release_data() return.

Since the recorded contents of the changeset are not used thereafter, just
don't pass it.

Found by Linux Verification Center (linuxtesting.org) with Syzkaller.

Reported-by: [email protected]
Closes: https://lore.kernel.org/lkml/[email protected]
Fixes: af0e2aab3b70 ("btrfs: qgroup: flush reservations during quota disable")
CC: [email protected] # 6.10+
Reviewed-by: Boris Burkov <[email protected]>
Reviewed-by: Qu Wenruo <[email protected]>
Signed-off-by: Fedor Pchelkin <[email protected]>
Signed-off-by: David Sterba <[email protected]>
6 months agohwmon: (hp-wmi-sensors) Check if WMI event data exists
Armin Wolf [Sun, 1 Sep 2024 03:10:51 +0000 (05:10 +0200)]
hwmon: (hp-wmi-sensors) Check if WMI event data exists

The BIOS can choose to return no event data in response to a
WMI event, so the ACPI object passed to the WMI notify handler
can be NULL.

Check for such a situation and ignore the event in such a case.

Fixes: 23902f98f8d4 ("hwmon: add HP WMI Sensors driver")
Signed-off-by: Armin Wolf <[email protected]>
Reviewed-by: Ilpo Järvinen <[email protected]>
Message-ID: <20240901031055[email protected]>
Signed-off-by: Guenter Roeck <[email protected]>
6 months agoMAINTAINERS: Remove Wedson as Rust maintainer
Wedson Almeida Filho [Wed, 28 Aug 2024 21:11:17 +0000 (18:11 -0300)]
MAINTAINERS: Remove Wedson as Rust maintainer

I am retiring from the project, so removing myself from MAINTAINERS as I won't
have time to dedicate to it.

Signed-off-by: Wedson Almeida Filho <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Miguel Ojeda <[email protected]>
6 months agorust: macros: provide correct provenance when constructing THIS_MODULE
Boqun Feng [Wed, 28 Aug 2024 18:01:29 +0000 (11:01 -0700)]
rust: macros: provide correct provenance when constructing THIS_MODULE

Currently while defining `THIS_MODULE` symbol in `module!()`, the
pointer used to construct `ThisModule` is derived from an immutable
reference of `__this_module`, which means the pointer doesn't have
the provenance for writing, and that means any write to that pointer
is UB regardless of data races or not. However, the usage of
`THIS_MODULE` includes passing this pointer to functions that may write
to it (probably in unsafe code), and this will create soundness issues.

One way to fix this is using `addr_of_mut!()` but that requires the
unstable feature "const_mut_refs". So instead of `addr_of_mut()!`,
an extern static `Opaque` is used here: since `Opaque<T>` is transparent
to `T`, an extern static `Opaque` will just wrap the C symbol (defined
in a C compile unit) in an `Opaque`, which provides a pointer with
writable provenance via `Opaque::get()`. This fix the potential UBs
because of pointer provenance unmatched.

Reported-by: Alice Ryhl <[email protected]>
Signed-off-by: Boqun Feng <[email protected]>
Reviewed-by: Alice Ryhl <[email protected]>
Reviewed-by: Trevor Gross <[email protected]>
Reviewed-by: Benno Lossin <[email protected]>
Reviewed-by: Gary Guo <[email protected]>
Closes: https://rust-for-linux.zulipchat.com/#narrow/stream/x/topic/x/near/465412664
Fixes: 1fbde52bde73 ("rust: add `macros` crate")
Cc: [email protected] # 6.6.x: be2ca1e03965: ("rust: types: Make Opaque::get const")
Link: https://lore.kernel.org/r/[email protected]
[ Fixed two typos, reworded title. - Miguel ]
Signed-off-by: Miguel Ojeda <[email protected]>
6 months agoMerge tag 'ata-6.11-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/libata...
Linus Torvalds [Mon, 2 Sep 2024 02:59:59 +0000 (19:59 -0700)]
Merge tag 'ata-6.11-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/libata/linux

Pull ata fix from Damien Le Moal:

 - Fix a potential memory leak in the ata host initialization code (from
   Zheng)

* tag 'ata-6.11-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/libata/linux:
  ata: libata: Fix memory leak for error path in ata_host_alloc()

6 months agoalloc_tag: fix allocation tag reporting when CONFIG_MODULES=n
Suren Baghdasaryan [Wed, 28 Aug 2024 23:15:36 +0000 (16:15 -0700)]
alloc_tag: fix allocation tag reporting when CONFIG_MODULES=n

codetag_module_init() is used to initialize sections containing allocation
tags.  This function is used to initialize module sections as well as core
kernel sections, in which case the module parameter is set to NULL.  This
function has to be called even when CONFIG_MODULES=n to initialize core
kernel allocation tag sections.  When CONFIG_MODULES=n, this function is a
NOP, which is wrong.  This leads to /proc/allocinfo reported as empty.
Fix this by making it independent of CONFIG_MODULES.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 916cc5167cc6 ("lib: code tagging framework")
Signed-off-by: Suren Baghdasaryan <[email protected]>
Cc: David Hildenbrand <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Kent Overstreet <[email protected]>
Cc: Pasha Tatashin <[email protected]>
Cc: Sourav Panda <[email protected]>
Cc: Vlastimil Babka <[email protected]>
Cc: <[email protected]> [6.10+]
Signed-off-by: Andrew Morton <[email protected]>
6 months agomm: vmalloc: optimize vmap_lazy_nr arithmetic when purging each vmap_area
Adrian Huang [Thu, 29 Aug 2024 13:06:33 +0000 (21:06 +0800)]
mm: vmalloc: optimize vmap_lazy_nr arithmetic when purging each vmap_area

When running the vmalloc stress on a 448-core system, observe the average
latency of purge_vmap_node() is about 2 seconds by using the eBPF/bcc
'funclatency.py' tool [1].

  # /your-git-repo/bcc/tools/funclatency.py -u purge_vmap_node & pid1=$! && sleep 8 && modprobe test_vmalloc nr_threads=$(nproc) run_test_mask=0x7; kill -SIGINT $pid1

     usecs             : count    distribution
        0 -> 1         : 0       |                                        |
        2 -> 3         : 29      |                                        |
        4 -> 7         : 19      |                                        |
        8 -> 15        : 56      |                                        |
       16 -> 31        : 483     |****                                    |
       32 -> 63        : 1548    |************                            |
       64 -> 127       : 2634    |*********************                   |
      128 -> 255       : 2535    |*********************                   |
      256 -> 511       : 1776    |**************                          |
      512 -> 1023      : 1015    |********                                |
     1024 -> 2047      : 573     |****                                    |
     2048 -> 4095      : 488     |****                                    |
     4096 -> 8191      : 1091    |*********                               |
     8192 -> 16383     : 3078    |*************************               |
    16384 -> 32767     : 4821    |****************************************|
    32768 -> 65535     : 3318    |***************************             |
    65536 -> 131071    : 1718    |**************                          |
   131072 -> 262143    : 2220    |******************                      |
   262144 -> 524287    : 1147    |*********                               |
   524288 -> 1048575   : 1179    |*********                               |
  1048576 -> 2097151   : 822     |******                                  |
  2097152 -> 4194303   : 906     |*******                                 |
  4194304 -> 8388607   : 2148    |*****************                       |
  8388608 -> 16777215  : 4497    |*************************************   |
 16777216 -> 33554431  : 289     |**                                      |

  avg = 2041714 usecs, total: 78381401772 usecs, count: 38390

  The worst case is over 16-33 seconds, so soft lockup is triggered [2].

[Root Cause]
1) Each purge_list has the long list. The following shows the number of
   vmap_area is purged.

   crash> p vmap_nodes
   vmap_nodes = $27 = (struct vmap_node *) 0xff2de5a900100000
   crash> vmap_node 0xff2de5a900100000 128 | grep nr_purged
     nr_purged = 663070
     ...
     nr_purged = 821670
     nr_purged = 692214
     nr_purged = 726808
     ...

2) atomic_long_sub() employs the 'lock' prefix to ensure the atomic
   operation when purging each vmap_area. However, the iteration is over
   600000 vmap_area (See 'nr_purged' above).

   Here is objdump output:

     $ objdump -D vmlinux
     ffffffff813e8c80 <purge_vmap_node>:
     ...
     ffffffff813e8d70:  f0 48 29 2d 68 0c bb  lock sub %rbp,0x2bb0c68(%rip)
     ...

   Quote from "Instruction tables" pdf file [3]:
     Instructions with a LOCK prefix have a long latency that depends on
     cache organization and possibly RAM speed. If there are multiple
     processors or cores or direct memory access (DMA) devices, then all
     locked instructions will lock a cache line for exclusive access,
     which may involve RAM access. A LOCK prefix typically costs more
     than a hundred clock cycles, even on single-processor systems.

   That's why the latency of purge_vmap_node() dramatically increases
   on a many-core system: One core is busy on purging each vmap_area of
   the *long* purge_list and executing atomic_long_sub() for each
   vmap_area, while other cores free vmalloc allocations and execute
   atomic_long_add_return() in free_vmap_area_noflush().

[Solution]
Employ a local variable to record the total purged pages, and execute
atomic_long_sub() after the traversal of the purge_list is done. The
experiment result shows the latency improvement is 99%.

[Experiment Result]
1) System Configuration: Three servers (with HT-enabled) are tested.
     * 72-core server: 3rd Gen Intel Xeon Scalable Processor*1
     * 192-core server: 5th Gen Intel Xeon Scalable Processor*2
     * 448-core server: AMD Zen 4 Processor*2

2) Kernel Config
     * CONFIG_KASAN is disabled

3) The data in column "w/o patch" and "w/ patch"
     * Unit: micro seconds (us)
     * Each data is the average of 3-time measurements

         System        w/o patch (us)   w/ patch (us)    Improvement (%)
     ---------------   --------------   -------------    -------------
     72-core server          2194              14            99.36%
     192-core server       143799            1139            99.21%
     448-core server      1992122            6883            99.65%

[1] https://github.com/iovisor/bcc/blob/master/tools/funclatency.py
[2] https://gist.github.com/AdrianHuang/37c15f67b45407b83c2d32f918656c12
[3] https://www.agner.org/optimize/instruction_tables.pdf

Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Adrian Huang <[email protected]>
Reviewed-by: Uladzislau Rezki (Sony) <[email protected]>
Cc: Christoph Hellwig <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agomailmap: update entry for Jan Kuliga
Jan Kuliga [Fri, 30 Aug 2024 09:56:58 +0000 (11:56 +0200)]
mailmap: update entry for Jan Kuliga

Soon I won't be able to use my current email address.

Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Jan Kuliga <[email protected]>
Cc: David S. Miller <[email protected]>
Cc: Matthieu Baerts (NGI0) <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agocodetag: debug: mark codetags for poisoned page as empty
Hao Ge [Sun, 25 Aug 2024 16:36:49 +0000 (00:36 +0800)]
codetag: debug: mark codetags for poisoned page as empty

When PG_hwpoison pages are freed they are treated differently in
free_pages_prepare() and instead of being released they are isolated.

Page allocation tag counters are decremented at this point since the page
is considered not in use.  Later on when such pages are released by
unpoison_memory(), the allocation tag counters will be decremented again
and the following warning gets reported:

[  113.930443][ T3282] ------------[ cut here ]------------
[  113.931105][ T3282] alloc_tag was not set
[  113.931576][ T3282] WARNING: CPU: 2 PID: 3282 at ./include/linux/alloc_tag.h:130 pgalloc_tag_sub.part.66+0x154/0x164
[  113.932866][ T3282] Modules linked in: hwpoison_inject fuse ip6t_rpfilter ip6t_REJECT nf_reject_ipv6 ipt_REJECT nf_reject_ipv4 xt_conntrack ebtable_nat ebtable_broute ip6table_nat ip6table_man4
[  113.941638][ T3282] CPU: 2 UID: 0 PID: 3282 Comm: madvise11 Kdump: loaded Tainted: G        W          6.11.0-rc4-dirty #18
[  113.943003][ T3282] Tainted: [W]=WARN
[  113.943453][ T3282] Hardware name: QEMU KVM Virtual Machine, BIOS unknown 2/2/2022
[  113.944378][ T3282] pstate: 40400005 (nZcv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[  113.945319][ T3282] pc : pgalloc_tag_sub.part.66+0x154/0x164
[  113.946016][ T3282] lr : pgalloc_tag_sub.part.66+0x154/0x164
[  113.946706][ T3282] sp : ffff800087093a10
[  113.947197][ T3282] x29: ffff800087093a10 x28: ffff0000d7a9d400 x27: ffff80008249f0a0
[  113.948165][ T3282] x26: 0000000000000000 x25: ffff80008249f2b0 x24: 0000000000000000
[  113.949134][ T3282] x23: 0000000000000001 x22: 0000000000000001 x21: 0000000000000000
[  113.950597][ T3282] x20: ffff0000c08fcad8 x19: ffff80008251e000 x18: ffffffffffffffff
[  113.952207][ T3282] x17: 0000000000000000 x16: 0000000000000000 x15: ffff800081746210
[  113.953161][ T3282] x14: 0000000000000000 x13: 205d323832335420 x12: 5b5d353031313339
[  113.954120][ T3282] x11: ffff800087093500 x10: 000000000000005d x9 : 00000000ffffffd0
[  113.955078][ T3282] x8 : 7f7f7f7f7f7f7f7f x7 : ffff80008236ba90 x6 : c0000000ffff7fff
[  113.956036][ T3282] x5 : ffff000b34bf4dc8 x4 : ffff8000820aba90 x3 : 0000000000000001
[  113.956994][ T3282] x2 : ffff800ab320f000 x1 : 841d1e35ac932e00 x0 : 0000000000000000
[  113.957962][ T3282] Call trace:
[  113.958350][ T3282]  pgalloc_tag_sub.part.66+0x154/0x164
[  113.959000][ T3282]  pgalloc_tag_sub+0x14/0x1c
[  113.959539][ T3282]  free_unref_page+0xf4/0x4b8
[  113.960096][ T3282]  __folio_put+0xd4/0x120
[  113.960614][ T3282]  folio_put+0x24/0x50
[  113.961103][ T3282]  unpoison_memory+0x4f0/0x5b0
[  113.961678][ T3282]  hwpoison_unpoison+0x30/0x48 [hwpoison_inject]
[  113.962436][ T3282]  simple_attr_write_xsigned.isra.34+0xec/0x1cc
[  113.963183][ T3282]  simple_attr_write+0x38/0x48
[  113.963750][ T3282]  debugfs_attr_write+0x54/0x80
[  113.964330][ T3282]  full_proxy_write+0x68/0x98
[  113.964880][ T3282]  vfs_write+0xdc/0x4d0
[  113.965372][ T3282]  ksys_write+0x78/0x100
[  113.965875][ T3282]  __arm64_sys_write+0x24/0x30
[  113.966440][ T3282]  invoke_syscall+0x7c/0x104
[  113.966984][ T3282]  el0_svc_common.constprop.1+0x88/0x104
[  113.967652][ T3282]  do_el0_svc+0x2c/0x38
[  113.968893][ T3282]  el0_svc+0x3c/0x1b8
[  113.969379][ T3282]  el0t_64_sync_handler+0x98/0xbc
[  113.969980][ T3282]  el0t_64_sync+0x19c/0x1a0
[  113.970511][ T3282] ---[ end trace 0000000000000000 ]---

To fix this, clear the page tag reference after the page got isolated
and accounted for.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: d224eb0287fb ("codetag: debug: mark codetags for reserved pages as empty")
Signed-off-by: Hao Ge <[email protected]>
Reviewed-by: Miaohe Lin <[email protected]>
Acked-by: Suren Baghdasaryan <[email protected]>
Cc: David Hildenbrand <[email protected]>
Cc: Hao Ge <[email protected]>
Cc: Kent Overstreet <[email protected]>
Cc: Naoya Horiguchi <[email protected]>
Cc: Pasha Tatashin <[email protected]>
Cc: <[email protected]> [6.10+]
Signed-off-by: Andrew Morton <[email protected]>
6 months agomm/memcontrol: respect zswap.writeback setting from parent cg too
Mike Yuan [Fri, 23 Aug 2024 16:27:06 +0000 (16:27 +0000)]
mm/memcontrol: respect zswap.writeback setting from parent cg too

Currently, the behavior of zswap.writeback wrt.  the cgroup hierarchy
seems a bit odd.  Unlike zswap.max, it doesn't honor the value from parent
cgroups.  This surfaced when people tried to globally disable zswap
writeback, i.e.  reserve physical swap space only for hibernation [1] -
disabling zswap.writeback only for the root cgroup results in subcgroups
with zswap.writeback=1 still performing writeback.

The inconsistency became more noticeable after I introduced the
MemoryZSwapWriteback= systemd unit setting [2] for controlling the knob.
The patch assumed that the kernel would enforce the value of parent
cgroups.  It could probably be workarounded from systemd's side, by going
up the slice unit tree and inheriting the value.  Yet I think it's more
sensible to make it behave consistently with zswap.max and friends.

[1] https://wiki.archlinux.org/title/Power_management/Suspend_and_hibernate#Disable_zswap_writeback_to_use_the_swap_space_only_for_hibernation
[2] https://github.com/systemd/systemd/pull/31734

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 501a06fe8e4c ("zswap: memcontrol: implement zswap writeback disabling")
Signed-off-by: Mike Yuan <[email protected]>
Reviewed-by: Nhat Pham <[email protected]>
Acked-by: Yosry Ahmed <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: Michal Koutný <[email protected]>
Cc: Muchun Song <[email protected]>
Cc: Roman Gushchin <[email protected]>
Cc: Shakeel Butt <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agoscripts: fix gfp-translate after ___GFP_*_BITS conversion to an enum
Marc Zyngier [Fri, 23 Aug 2024 16:38:50 +0000 (17:38 +0100)]
scripts: fix gfp-translate after ___GFP_*_BITS conversion to an enum

Richard reports that since 772dd0342727c ("mm: enumerate all gfp flags"),
gfp-translate is broken, as the bit numbers are implicit, leaving the
shell script unable to extract them.  Even more, some bits are now at a
variable location, making it double extra hard to parse using a simple
shell script.

Use a brute-force approach to the problem by generating a small C stub
that will use the enum to dump the interesting bits.

As an added bonus, we are now able to identify invalid bits for a given
configuration.  As an added drawback, we cannot parse include files that
predate this change anymore.  Tough luck.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 772dd0342727 ("mm: enumerate all gfp flags")
Signed-off-by: Marc Zyngier <[email protected]>
Reported-by: Richard Weinberger <[email protected]>
Cc: Petr Tesařík <[email protected]>
Cc: Suren Baghdasaryan <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agoRevert "mm: skip CMA pages when they are not available"
Usama Arif [Wed, 21 Aug 2024 19:26:07 +0000 (20:26 +0100)]
Revert "mm: skip CMA pages when they are not available"

This reverts commit 5da226dbfce3 ("mm: skip CMA pages when they are not
available") and b7108d66318a ("Multi-gen LRU: skip CMA pages when they are
not eligible").

lruvec->lru_lock is highly contended and is held when calling
isolate_lru_folios.  If the lru has a large number of CMA folios
consecutively, while the allocation type requested is not MIGRATE_MOVABLE,
isolate_lru_folios can hold the lock for a very long time while it skips
those.  For FIO workload, ~150million order=0 folios were skipped to
isolate a few ZONE_DMA folios [1].  This can cause lockups [1] and high
memory pressure for extended periods of time [2].

Remove skipping CMA for MGLRU as well, as it was introduced in sort_folio
for the same resaon as 5da226dbfce3a2f44978c2c7cf88166e69a6788b.

[1] https://lore.kernel.org/all/CAOUHufbkhMZYz20aM_3rHZ3OcK4m2puji2FGpUpn_-DevGk3Kg@mail.gmail.com/
[2] https://lore.kernel.org/all/[email protected]/

[[email protected]: also revert b7108d66318a, per Johannes]
Link: https://lkml.kernel.org/r/[email protected]
Link: https://lkml.kernel.org/r/[email protected]
Link: https://lkml.kernel.org/r/[email protected]
Fixes: 5da226dbfce3 ("mm: skip CMA pages when they are not available")
Signed-off-by: Usama Arif <[email protected]>
Acked-by: Johannes Weiner <[email protected]>
Cc: Bharata B Rao <[email protected]>
Cc: Breno Leitao <[email protected]>
Cc: David Hildenbrand <[email protected]>
Cc: Matthew Wilcox <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Vlastimil Babka <[email protected]>
Cc: Yu Zhao <[email protected]>
Cc: Zhaoyang Huang <[email protected]>
Cc: Zhaoyang Huang <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agomaple_tree: remove rcu_read_lock() from mt_validate()
Liam R. Howlett [Tue, 20 Aug 2024 17:54:17 +0000 (13:54 -0400)]
maple_tree: remove rcu_read_lock() from mt_validate()

The write lock should be held when validating the tree to avoid updates
racing with checks.  Holding the rcu read lock during a large tree
validation may also cause a prolonged rcu read window and "rcu_preempt
detected stalls" warnings.

Link: https://lore.kernel.org/all/[email protected]/
Link: https://lkml.kernel.org/r/[email protected]
Fixes: 54a611b60590 ("Maple Tree: add new data structure")
Signed-off-by: Liam R. Howlett <[email protected]>
Reported-by: [email protected]
Cc: Hillf Danton <[email protected]>
Cc: Matthew Wilcox <[email protected]>
Cc: "Paul E. McKenney" <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agokexec_file: fix elfcorehdr digest exclusion when CONFIG_CRASH_HOTPLUG=y
Petr Tesarik [Mon, 5 Aug 2024 15:07:50 +0000 (17:07 +0200)]
kexec_file: fix elfcorehdr digest exclusion when CONFIG_CRASH_HOTPLUG=y

Fix the condition to exclude the elfcorehdr segment from the SHA digest
calculation.

The j iterator is an index into the output sha_regions[] array, not into
the input image->segment[] array.  Once it reaches
image->elfcorehdr_index, all subsequent segments are excluded.  Besides,
if the purgatory segment precedes the elfcorehdr segment, the elfcorehdr
may be wrongly included in the calculation.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: f7cc804a9fd4 ("kexec: exclude elfcorehdr from the segment digest")
Signed-off-by: Petr Tesarik <[email protected]>
Acked-by: Baoquan He <[email protected]>
Cc: Eric Biederman <[email protected]>
Cc: Hari Bathini <[email protected]>
Cc: Sourabh Jain <[email protected]>
Cc: Eric DeVolder <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agomm/slub: add check for s->flags in the alloc_tagging_slab_free_hook
Hao Ge [Fri, 16 Aug 2024 01:33:36 +0000 (09:33 +0800)]
mm/slub: add check for s->flags in the alloc_tagging_slab_free_hook

When enable CONFIG_MEMCG & CONFIG_KFENCE & CONFIG_KMEMLEAK, the following
warning always occurs,This is because the following call stack occurred:
mem_pool_alloc
    kmem_cache_alloc_noprof
        slab_alloc_node
            kfence_alloc

Once the kfence allocation is successful,slab->obj_exts will not be empty,
because it has already been assigned a value in kfence_init_pool.

Since in the prepare_slab_obj_exts_hook function,we perform a check for
s->flags & (SLAB_NO_OBJ_EXT | SLAB_NOLEAKTRACE),the alloc_tag_add function
will not be called as a result.Therefore,ref->ct remains NULL.

However,when we call mem_pool_free,since obj_ext is not empty, it
eventually leads to the alloc_tag_sub scenario being invoked.  This is
where the warning occurs.

So we should add corresponding checks in the alloc_tagging_slab_free_hook.
For __GFP_NO_OBJ_EXT case,I didn't see the specific case where it's using
kfence,so I won't add the corresponding check in
alloc_tagging_slab_free_hook for now.

[    3.734349] ------------[ cut here ]------------
[    3.734807] alloc_tag was not set
[    3.735129] WARNING: CPU: 4 PID: 40 at ./include/linux/alloc_tag.h:130 kmem_cache_free+0x444/0x574
[    3.735866] Modules linked in: autofs4
[    3.736211] CPU: 4 UID: 0 PID: 40 Comm: ksoftirqd/4 Tainted: G        W          6.11.0-rc3-dirty #1
[    3.736969] Tainted: [W]=WARN
[    3.737258] Hardware name: QEMU KVM Virtual Machine, BIOS unknown 2/2/2022
[    3.737875] pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[    3.738501] pc : kmem_cache_free+0x444/0x574
[    3.738951] lr : kmem_cache_free+0x444/0x574
[    3.739361] sp : ffff80008357bb60
[    3.739693] x29: ffff80008357bb70 x28: 0000000000000000 x27: 0000000000000000
[    3.740338] x26: ffff80008207f000 x25: ffff000b2eb2fd60 x24: ffff0000c0005700
[    3.740982] x23: ffff8000804229e4 x22: ffff800082080000 x21: ffff800081756000
[    3.741630] x20: fffffd7ff8253360 x19: 00000000000000a8 x18: ffffffffffffffff
[    3.742274] x17: ffff800ab327f000 x16: ffff800083398000 x15: ffff800081756df0
[    3.742919] x14: 0000000000000000 x13: 205d344320202020 x12: 5b5d373038343337
[    3.743560] x11: ffff80008357b650 x10: 000000000000005d x9 : 00000000ffffffd0
[    3.744231] x8 : 7f7f7f7f7f7f7f7f x7 : ffff80008237bad0 x6 : c0000000ffff7fff
[    3.744907] x5 : ffff80008237ba78 x4 : ffff8000820bbad0 x3 : 0000000000000001
[    3.745580] x2 : 68d66547c09f7800 x1 : 68d66547c09f7800 x0 : 0000000000000000
[    3.746255] Call trace:
[    3.746530]  kmem_cache_free+0x444/0x574
[    3.746931]  mem_pool_free+0x44/0xf4
[    3.747306]  free_object_rcu+0xc8/0xdc
[    3.747693]  rcu_do_batch+0x234/0x8a4
[    3.748075]  rcu_core+0x230/0x3e4
[    3.748424]  rcu_core_si+0x14/0x1c
[    3.748780]  handle_softirqs+0x134/0x378
[    3.749189]  run_ksoftirqd+0x70/0x9c
[    3.749560]  smpboot_thread_fn+0x148/0x22c
[    3.749978]  kthread+0x10c/0x118
[    3.750323]  ret_from_fork+0x10/0x20
[    3.750696] ---[ end trace 0000000000000000 ]---

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 4b8736964640 ("mm/slab: add allocation accounting into slab allocation and free paths")
Signed-off-by: Hao Ge <[email protected]>
Cc: Christoph Lameter <[email protected]>
Cc: David Rientjes <[email protected]>
Cc: Hyeonggon Yoo <[email protected]>
Cc: Joonsoo Kim <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Kent Overstreet <[email protected]>
Cc: Pekka Enberg <[email protected]>
Cc: Roman Gushchin <[email protected]>
Cc: Suren Baghdasaryan <[email protected]>
Cc: Vlastimil Babka <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agonilfs2: fix state management in error path of log writing function
Ryusuke Konishi [Wed, 14 Aug 2024 10:11:19 +0000 (19:11 +0900)]
nilfs2: fix state management in error path of log writing function

After commit a694291a6211 ("nilfs2: separate wait function from
nilfs_segctor_write") was applied, the log writing function
nilfs_segctor_do_construct() was able to issue I/O requests continuously
even if user data blocks were split into multiple logs across segments,
but two potential flaws were introduced in its error handling.

First, if nilfs_segctor_begin_construction() fails while creating the
second or subsequent logs, the log writing function returns without
calling nilfs_segctor_abort_construction(), so the writeback flag set on
pages/folios will remain uncleared.  This causes page cache operations to
hang waiting for the writeback flag.  For example,
truncate_inode_pages_final(), which is called via nilfs_evict_inode() when
an inode is evicted from memory, will hang.

Second, the NILFS_I_COLLECTED flag set on normal inodes remain uncleared.
As a result, if the next log write involves checkpoint creation, that's
fine, but if a partial log write is performed that does not, inodes with
NILFS_I_COLLECTED set are erroneously removed from the "sc_dirty_files"
list, and their data and b-tree blocks may not be written to the device,
corrupting the block mapping.

Fix these issues by uniformly calling nilfs_segctor_abort_construction()
on failure of each step in the loop in nilfs_segctor_do_construct(),
having it clean up logs and segment usages according to progress, and
correcting the conditions for calling nilfs_redirty_inodes() to ensure
that the NILFS_I_COLLECTED flag is cleared.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: a694291a6211 ("nilfs2: separate wait function from nilfs_segctor_write")
Signed-off-by: Ryusuke Konishi <[email protected]>
Tested-by: Ryusuke Konishi <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agonilfs2: fix missing cleanup on rollforward recovery error
Ryusuke Konishi [Sat, 10 Aug 2024 06:52:42 +0000 (15:52 +0900)]
nilfs2: fix missing cleanup on rollforward recovery error

In an error injection test of a routine for mount-time recovery, KASAN
found a use-after-free bug.

It turned out that if data recovery was performed using partial logs
created by dsync writes, but an error occurred before starting the log
writer to create a recovered checkpoint, the inodes whose data had been
recovered were left in the ns_dirty_files list of the nilfs object and
were not freed.

Fix this issue by cleaning up inodes that have read the recovery data if
the recovery routine fails midway before the log writer starts.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 0f3e1c7f23f8 ("nilfs2: recovery functions")
Signed-off-by: Ryusuke Konishi <[email protected]>
Tested-by: Ryusuke Konishi <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agonilfs2: protect references to superblock parameters exposed in sysfs
Ryusuke Konishi [Sun, 11 Aug 2024 10:03:20 +0000 (19:03 +0900)]
nilfs2: protect references to superblock parameters exposed in sysfs

The superblock buffers of nilfs2 can not only be overwritten at runtime
for modifications/repairs, but they are also regularly swapped, replaced
during resizing, and even abandoned when degrading to one side due to
backing device issues.  So, accessing them requires mutual exclusion using
the reader/writer semaphore "nilfs->ns_sem".

Some sysfs attribute show methods read this superblock buffer without the
necessary mutual exclusion, which can cause problems with pointer
dereferencing and memory access, so fix it.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: da7141fb78db ("nilfs2: add /sys/fs/nilfs2/<device> group")
Signed-off-by: Ryusuke Konishi <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agouserfaultfd: don't BUG_ON() if khugepaged yanks our page table
Jann Horn [Tue, 13 Aug 2024 20:25:22 +0000 (22:25 +0200)]
userfaultfd: don't BUG_ON() if khugepaged yanks our page table

Since khugepaged was changed to allow retracting page tables in file
mappings without holding the mmap lock, these BUG_ON()s are wrong - get
rid of them.

We could also remove the preceding "if (unlikely(...))" block, but then we
could reach pte_offset_map_lock() with transhuge pages not just for file
mappings but also for anonymous mappings - which would probably be fine
but I think is not necessarily expected.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 1d65b771bc08 ("mm/khugepaged: retract_page_tables() without mmap or vma lock")
Signed-off-by: Jann Horn <[email protected]>
Reviewed-by: Qi Zheng <[email protected]>
Acked-by: David Hildenbrand <[email protected]>
Cc: Andrea Arcangeli <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Pavel Emelyanov <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agouserfaultfd: fix checks for huge PMDs
Jann Horn [Tue, 13 Aug 2024 20:25:21 +0000 (22:25 +0200)]
userfaultfd: fix checks for huge PMDs

Patch series "userfaultfd: fix races around pmd_trans_huge() check", v2.

The pmd_trans_huge() code in mfill_atomic() is wrong in three different
ways depending on kernel version:

1. The pmd_trans_huge() check is racy and can lead to a BUG_ON() (if you hit
   the right two race windows) - I've tested this in a kernel build with
   some extra mdelay() calls. See the commit message for a description
   of the race scenario.
   On older kernels (before 6.5), I think the same bug can even
   theoretically lead to accessing transhuge page contents as a page table
   if you hit the right 5 narrow race windows (I haven't tested this case).
2. As pointed out by Qi Zheng, pmd_trans_huge() is not sufficient for
   detecting PMDs that don't point to page tables.
   On older kernels (before 6.5), you'd just have to win a single fairly
   wide race to hit this.
   I've tested this on 6.1 stable by racing migration (with a mdelay()
   patched into try_to_migrate()) against UFFDIO_ZEROPAGE - on my x86
   VM, that causes a kernel oops in ptlock_ptr().
3. On newer kernels (>=6.5), for shmem mappings, khugepaged is allowed
   to yank page tables out from under us (though I haven't tested that),
   so I think the BUG_ON() checks in mfill_atomic() are just wrong.

I decided to write two separate fixes for these (one fix for bugs 1+2, one
fix for bug 3), so that the first fix can be backported to kernels
affected by bugs 1+2.

This patch (of 2):

This fixes two issues.

I discovered that the following race can occur:

  mfill_atomic                other thread
  ============                ============
                              <zap PMD>
  pmdp_get_lockless() [reads none pmd]
  <bail if trans_huge>
  <if none:>
                              <pagefault creates transhuge zeropage>
    __pte_alloc [no-op]
                              <zap PMD>
  <bail if pmd_trans_huge(*dst_pmd)>
  BUG_ON(pmd_none(*dst_pmd))

I have experimentally verified this in a kernel with extra mdelay() calls;
the BUG_ON(pmd_none(*dst_pmd)) triggers.

On kernels newer than commit 0d940a9b270b ("mm/pgtable: allow
pte_offset_map[_lock]() to fail"), this can't lead to anything worse than
a BUG_ON(), since the page table access helpers are actually designed to
deal with page tables concurrently disappearing; but on older kernels
(<=6.4), I think we could probably theoretically race past the two
BUG_ON() checks and end up treating a hugepage as a page table.

The second issue is that, as Qi Zheng pointed out, there are other types
of huge PMDs that pmd_trans_huge() can't catch: devmap PMDs and swap PMDs
(in particular, migration PMDs).

On <=6.4, this is worse than the first issue: If mfill_atomic() runs on a
PMD that contains a migration entry (which just requires winning a single,
fairly wide race), it will pass the PMD to pte_offset_map_lock(), which
assumes that the PMD points to a page table.

Breakage follows: First, the kernel tries to take the PTE lock (which will
crash or maybe worse if there is no "struct page" for the address bits in
the migration entry PMD - I think at least on X86 there usually is no
corresponding "struct page" thanks to the PTE inversion mitigation, amd64
looks different).

If that didn't crash, the kernel would next try to write a PTE into what
it wrongly thinks is a page table.

As part of fixing these issues, get rid of the check for pmd_trans_huge()
before __pte_alloc() - that's redundant, we're going to have to check for
that after the __pte_alloc() anyway.

Backport note: pmdp_get_lockless() is pmd_read_atomic() in older kernels.

Link: https://lkml.kernel.org/r/[email protected]
Link: https://lkml.kernel.org/r/[email protected]
Fixes: c1a4de99fada ("userfaultfd: mcopy_atomic|mfill_zeropage: UFFDIO_COPY|UFFDIO_ZEROPAGE preparation")
Signed-off-by: Jann Horn <[email protected]>
Acked-by: David Hildenbrand <[email protected]>
Cc: Andrea Arcangeli <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Jann Horn <[email protected]>
Cc: Pavel Emelyanov <[email protected]>
Cc: Qi Zheng <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agomm: vmalloc: ensure vmap_block is initialised before adding to queue
Will Deacon [Mon, 12 Aug 2024 17:16:06 +0000 (18:16 +0100)]
mm: vmalloc: ensure vmap_block is initialised before adding to queue

Commit 8c61291fd850 ("mm: fix incorrect vbq reference in
purge_fragmented_block") extended the 'vmap_block' structure to contain a
'cpu' field which is set at allocation time to the id of the initialising
CPU.

When a new 'vmap_block' is being instantiated by new_vmap_block(), the
partially initialised structure is added to the local 'vmap_block_queue'
xarray before the 'cpu' field has been initialised.  If another CPU is
concurrently walking the xarray (e.g.  via vm_unmap_aliases()), then it
may perform an out-of-bounds access to the remote queue thanks to an
uninitialised index.

This has been observed as UBSAN errors in Android:

 | Internal error: UBSAN: array index out of bounds: 00000000f2005512 [#1] PREEMPT SMP
 |
 | Call trace:
 |  purge_fragmented_block+0x204/0x21c
 |  _vm_unmap_aliases+0x170/0x378
 |  vm_unmap_aliases+0x1c/0x28
 |  change_memory_common+0x1dc/0x26c
 |  set_memory_ro+0x18/0x24
 |  module_enable_ro+0x98/0x238
 |  do_init_module+0x1b0/0x310

Move the initialisation of 'vb->cpu' in new_vmap_block() ahead of the
addition to the xarray.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 8c61291fd850 ("mm: fix incorrect vbq reference in purge_fragmented_block")
Signed-off-by: Will Deacon <[email protected]>
Reviewed-by: Baoquan He <[email protected]>
Reviewed-by: Uladzislau Rezki (Sony) <[email protected]>
Cc: Zhaoyang Huang <[email protected]>
Cc: Hailong.Liu <[email protected]>
Cc: Christoph Hellwig <[email protected]>
Cc: Lorenzo Stoakes <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agoselftests: mm: fix build errors on armhf
Muhammad Usama Anjum [Fri, 9 Aug 2024 08:25:11 +0000 (13:25 +0500)]
selftests: mm: fix build errors on armhf

The __NR_mmap isn't found on armhf.  The mmap() is commonly available
system call and its wrapper is present on all architectures.  So it should
be used directly.  It solves problem for armhf and doesn't create problem
for other architectures.

Remove sys_mmap() functions as they aren't doing anything else other than
calling mmap().  There is no need to set errno = 0 manually as glibc
always resets it.

For reference errors are as following:

  CC       seal_elf
seal_elf.c: In function 'sys_mmap':
seal_elf.c:39:33: error: '__NR_mmap' undeclared (first use in this function)
   39 |         sret = (void *) syscall(__NR_mmap, addr, len, prot,
      |                                 ^~~~~~~~~

mseal_test.c: In function 'sys_mmap':
mseal_test.c:90:33: error: '__NR_mmap' undeclared (first use in this function)
   90 |         sret = (void *) syscall(__NR_mmap, addr, len, prot,
      |                                 ^~~~~~~~~

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 4926c7a52de7 ("selftest mm/mseal memory sealing")
Signed-off-by: Muhammad Usama Anjum <[email protected]>
Cc: Jeff Xu <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Liam R. Howlett <[email protected]>
Cc: Shuah Khan <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
6 months agoMerge tag 'x86-urgent-2024-09-01' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 1 Sep 2024 21:43:08 +0000 (14:43 -0700)]
Merge tag 'x86-urgent-2024-09-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull x86 fixes from Thomas Gleixner:

 - x2apic_disable() clears x2apic_state and x2apic_mode unconditionally,
   even when the state is X2APIC_ON_LOCKED, which prevents the kernel to
   disable it thereby creating inconsistent state.

   Reorder the logic so it actually works correctly

 - The XSTATE logic for handling LBR is incorrect as it assumes that
   XSAVES supports LBR when the CPU supports LBR. In fact both
   conditions need to be true. Otherwise the enablement of LBR in the
   IA32_XSS MSR fails and subsequently the machine crashes on the next
   XRSTORS operation because IA32_XSS is not initialized.

   Cache the XSTATE support bit during init and make the related
   functions use this cached information and the LBR CPU feature bit to
   cure this.

 - Cure a long standing bug in KASLR

   KASLR uses the full address space between PAGE_OFFSET and vaddr_end
   to randomize the starting points of the direct map, vmalloc and
   vmemmap regions. It thereby limits the size of the direct map by
   using the installed memory size plus an extra configurable margin for
   hot-plug memory. This limitation is done to gain more randomization
   space because otherwise only the holes between the direct map,
   vmalloc, vmemmap and vaddr_end would be usable for randomizing.

   The limited direct map size is not exposed to the rest of the kernel,
   so the memory hot-plug and resource management related code paths
   still operate under the assumption that the available address space
   can be determined with MAX_PHYSMEM_BITS.

   request_free_mem_region() allocates from (1 << MAX_PHYSMEM_BITS) - 1
   downwards. That means the first allocation happens past the end of
   the direct map and if unlucky this address is in the vmalloc space,
   which causes high_memory to become greater than VMALLOC_START and
   consequently causes iounmap() to fail for valid ioremap addresses.

   Cure this by exposing the end of the direct map via PHYSMEM_END and
   use that for the memory hot-plug and resource management related
   places instead of relying on MAX_PHYSMEM_BITS. In the KASLR case
   PHYSMEM_END maps to a variable which is initialized by the KASLR
   initialization and otherwise it is based on MAX_PHYSMEM_BITS as
   before.

 - Prevent a data leak in mmio_read(). The TDVMCALL exposes the value of
   an initialized variabled on the stack to the VMM. The variable is
   only required as output value, so it does not have to exposed to the
   VMM in the first place.

 - Prevent an array overrun in the resource control code on systems with
   Sub-NUMA Clustering enabled because the code failed to adjust the
   index by the number of SNC nodes per L3 cache.

* tag 'x86-urgent-2024-09-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/resctrl: Fix arch_mbm_* array overrun on SNC
  x86/tdx: Fix data leak in mmio_read()
  x86/kaslr: Expose and use the end of the physical memory address space
  x86/fpu: Avoid writing LBR bit to IA32_XSS unless supported
  x86/apic: Make x2apic_disable() work correctly

6 months agoMerge tag 'perf-urgent-2024-09-01' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 1 Sep 2024 21:33:31 +0000 (14:33 -0700)]
Merge tag 'perf-urgent-2024-09-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull perf fix from Thomas Gleixner:
 "A single fix for x86 performance monitoring.

  Haswell PMUs suffer from several errata and require a limit the
  minimal period for counter events, otherwise they suffer from endless
  loops in the PMU interrupt"

* tag 'perf-urgent-2024-09-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf/x86/intel: Limit the period on Haswell

6 months agoMerge tag 'locking-urgent-2024-08-25' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 1 Sep 2024 21:26:33 +0000 (14:26 -0700)]
Merge tag 'locking-urgent-2024-08-25' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull locking fix from Thomas Gleixner:
 "A single fix for rt_mutex.

  The deadlock detection code drops into an infinite scheduling loop
  while still holding rt_mutex::wait_lock, which rightfully triggers a
  'scheduling in atomic' warning.

  Unlock it before that"

* tag 'locking-urgent-2024-08-25' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  rtmutex: Drop rt_mutex::wait_lock before scheduling

6 months agoMerge tag 'irq-urgent-2024-08-25' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 1 Sep 2024 21:19:00 +0000 (14:19 -0700)]
Merge tag 'irq-urgent-2024-08-25' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull irq fixes from Thomas Gleixner:
 "A set of fixes for interrupt chip drivers:

   - Unbreak the PLIC driver for Allwinner D1 systems

     The recent conversion of the PLIC driver to a platform driver broke
     Allwinnder D1 systems due to the deferred probing of platform
     drivers.

     Due to that the only timer available on D1 systems cannot get an
     interrupt, which causes the system to hang at boot. Other RISCV
     platforms are not affected because they provide the architected SBI
     timer which uses the built in core interrupt controller.

     Cure this by probing PLIC early on D1 systems

   - Cure a regression in ARM/GIC-V3 on 32-bit ARM systems caused by the
     recent addition of a initialization function, which accesses system
     registers before they are enabled. On 64-bit ARM they are enabled
     prior to that by sheer luck.

     Ensure they are enabled.

   - Cure a use before check problem in the MSI library. The existing
     NULL pointer check is too late.

   - Cure a lock order inversion in the ARM/GIC-V4 driver

   - Fix a IS_ERR() vs. NULL pointer check issue in the RISCV APLIC
     driver

   - Plug a reference count leak in the ARM/GIC-V2 driver"

* tag 'irq-urgent-2024-08-25' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip/irq-msi-lib: Check for NULL ops in msi_lib_irq_domain_select()
  irqchip/gic-v3: Init SRE before poking sysregs
  irqchip/gic-v2m: Fix refcount leak in gicv2m_of_init()
  irqchip/riscv-aplic: Fix an IS_ERR() vs NULL bug in probe()
  irqchip/gic-v4: Fix ordering between vmapp and vpe locks
  irqchip/sifive-plic: Probe plic driver early for Allwinner D1 platform

6 months agobcachefs: fix rebalance accounting
Kent Overstreet [Sun, 1 Sep 2024 19:53:03 +0000 (15:53 -0400)]
bcachefs: fix rebalance accounting

Fixes: 49aa7830396b ("bcachefs: Fix rebalance_work accounting")
Signed-off-by: Kent Overstreet <[email protected]>
6 months agohwmon: ltc2991: fix register bits defines
Pawel Dembicki [Fri, 30 Aug 2024 11:13:50 +0000 (13:13 +0200)]
hwmon: ltc2991: fix register bits defines

In the LTC2991, V5 and V6 channels use the low nibble of the
"V5, V6, V7, and V8 Control Register" for configuration, but currently,
the high nibble is defined.

This patch changes the defines to use the low nibble.

Fixes: 2b9ea4262ae9 ("hwmon: Add driver for ltc2991")
Signed-off-by: Pawel Dembicki <[email protected]>
Message-ID: <20240830111349[email protected]>
Signed-off-by: Guenter Roeck <[email protected]>
6 months agofscache: delete fscache_cookie_lru_timer when fscache exits to avoid UAF
Baokun Li [Mon, 26 Aug 2024 11:20:56 +0000 (19:20 +0800)]
fscache: delete fscache_cookie_lru_timer when fscache exits to avoid UAF

The fscache_cookie_lru_timer is initialized when the fscache module
is inserted, but is not deleted when the fscache module is removed.
If timer_reduce() is called before removing the fscache module,
the fscache_cookie_lru_timer will be added to the timer list of
the current cpu. Afterwards, a use-after-free will be triggered
in the softIRQ after removing the fscache module, as follows:

==================================================================
BUG: unable to handle page fault for address: fffffbfff803c9e9
 PF: supervisor read access in kernel mode
 PF: error_code(0x0000) - not-present page
PGD 21ffea067 P4D 21ffea067 PUD 21ffe6067 PMD 110a7c067 PTE 0
Oops: Oops: 0000 [#1] PREEMPT SMP KASAN PTI
CPU: 1 UID: 0 PID: 0 Comm: swapper/1 Tainted: G W 6.11.0-rc3 #855
Tainted: [W]=WARN
RIP: 0010:__run_timer_base.part.0+0x254/0x8a0
Call Trace:
 <IRQ>
 tmigr_handle_remote_up+0x627/0x810
 __walk_groups.isra.0+0x47/0x140
 tmigr_handle_remote+0x1fa/0x2f0
 handle_softirqs+0x180/0x590
 irq_exit_rcu+0x84/0xb0
 sysvec_apic_timer_interrupt+0x6e/0x90
 </IRQ>
 <TASK>
 asm_sysvec_apic_timer_interrupt+0x1a/0x20
RIP: 0010:default_idle+0xf/0x20
 default_idle_call+0x38/0x60
 do_idle+0x2b5/0x300
 cpu_startup_entry+0x54/0x60
 start_secondary+0x20d/0x280
 common_startup_64+0x13e/0x148
 </TASK>
Modules linked in: [last unloaded: netfs]
==================================================================

Therefore delete fscache_cookie_lru_timer when removing the fscahe module.

Fixes: 12bb21a29c19 ("fscache: Implement cookie user counting and resource pinning")
Cc: [email protected]
Signed-off-by: Baokun Li <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Acked-by: David Howells <[email protected]>
Signed-off-by: Christian Brauner <[email protected]>
6 months agoLinux 6.11-rc6 v6.11-rc6
Linus Torvalds [Sun, 1 Sep 2024 07:46:02 +0000 (19:46 +1200)]
Linux 6.11-rc6

6 months agoMerge tag 'v6.11-rc5-smb-client-fixes' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Sun, 1 Sep 2024 03:49:26 +0000 (15:49 +1200)]
Merge tag 'v6.11-rc5-smb-client-fixes' of git://git.samba.org/sfrench/cifs-2.6

Pull smb client fixes from Steve French:

 - copy_file_range fix

 - two read fixes including read past end of file rc fix and read retry
   crediting fix

 - falloc zero range fix

* tag 'v6.11-rc5-smb-client-fixes' of git://git.samba.org/sfrench/cifs-2.6:
  cifs: Fix FALLOC_FL_ZERO_RANGE to preflush buffered part of target region
  cifs: Fix copy offload to flush destination region
  netfs, cifs: Fix handling of short DIO read
  cifs: Fix lack of credit renegotiation on read retry

6 months agoMerge tag 'bcachefs-2024-08-21' of https://github.com/koverstreet/bcachefs
Linus Torvalds [Sun, 1 Sep 2024 03:23:20 +0000 (15:23 +1200)]
Merge tag 'bcachefs-2024-08-21' of https://github.com/koverstreet/bcachefs

Push bcachefs fixes from Kent Overstreet:
 "The data corruption in the buffered write path is troubling; inode
  lock should not have been able to cause that...

   - Fix a rare data corruption in the rebalance path, caught as a nonce
     inconsistency on encrypted filesystems

   - Revert lockless buffered write path

   - Mark more errors as autofix"

* tag 'bcachefs-2024-08-21' of https://github.com/koverstreet/bcachefs:
  bcachefs: Mark more errors as autofix
  bcachefs: Revert lockless buffered IO path
  bcachefs: Fix bch2_extents_match() false positive
  bcachefs: Fix failure to return error in data_update_index_update()

6 months agobcachefs: Mark more errors as autofix
Kent Overstreet [Thu, 22 Aug 2024 15:47:32 +0000 (11:47 -0400)]
bcachefs: Mark more errors as autofix

errors that are known to always be safe to fix should be autofix: this
should be most errors even at this point, but that will need some
thorough review.

note that errors are still logged in the superblock, so we'll still know
that they happened.

Signed-off-by: Kent Overstreet <[email protected]>
6 months agobcachefs: Revert lockless buffered IO path
Kent Overstreet [Sat, 31 Aug 2024 21:44:51 +0000 (17:44 -0400)]
bcachefs: Revert lockless buffered IO path

We had a report of data corruption on nixos when building installer
images.

https://github.com/NixOS/nixpkgs/pull/321055#issuecomment-2184131334

It seems that writes are being dropped, but only when issued by QEMU,
and possibly only in snapshot mode. It's undetermined if it's write
calls are being dropped or dirty folios.

Further testing, via minimizing the original patch to just the change
that skips the inode lock on non appends/truncates, reveals that it
really is just not taking the inode lock that causes the corruption: it
has nothing to do with the other logic changes for preserving write
atomicity in corner cases.

It's also kernel config dependent: it doesn't reproduce with the minimal
kernel config that ktest uses, but it does reproduce with nixos's distro
config. Bisection the kernel config initially pointer the finger at page
migration or compaction, but it appears that was erroneous; we haven't
yet determined what kernel config option actually triggers it.

Sadly it appears this will have to be reverted since we're getting too
close to release and my plate is full, but we'd _really_ like to fully
debug it.

My suspicion is that this patch is exposing a preexisting bug - the
inode lock actually covers very little in IO paths, and we have a
different lock (the pagecache add lock) that guards against races with
truncate here.

Fixes: 7e64c86cdc6c ("bcachefs: Buffered write path now can avoid the inode lock")
Signed-off-by: Kent Overstreet <[email protected]>
6 months agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux...
Linus Torvalds [Sat, 31 Aug 2024 21:18:48 +0000 (09:18 +1200)]
Merge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging

Pull misc fixes from Guenter Roeck.

These are fixes for regressions that Guenther has been reporting, and
the maintainers haven't picked up and sent in. With rc6 fairly imminent,
I'm taking them directly from Guenter.

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  apparmor: fix policy_unpack_test on big endian systems
  Revert "MIPS: csrc-r4k: Apply verification clocksource flags"
  microblaze: don't treat zero reserved memory regions as error

6 months agoMerge tag 'pwrseq-fixes-for-v6.11-rc6' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 31 Aug 2024 21:07:44 +0000 (09:07 +1200)]
Merge tag 'pwrseq-fixes-for-v6.11-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/brgl/linux

Pull power sequencing fix from Bartosz Golaszewski:
 "A follow-up fix for the power sequencing subsystem. It turned out the
  previous fix for this driver was incomplete and broke the WLAN support
  on some platforms. This addresses the issue.

   - set the direction of the wlan-enable GPIO to output after
     requesting it as-is"

* tag 'pwrseq-fixes-for-v6.11-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/brgl/linux:
  power: sequencing: qcom-wcn: set the wlan-enable GPIO to output

6 months agopower: sequencing: qcom-wcn: set the wlan-enable GPIO to output
Bartosz Golaszewski [Fri, 23 Aug 2024 11:55:00 +0000 (13:55 +0200)]
power: sequencing: qcom-wcn: set the wlan-enable GPIO to output

Commit a9aaf1ff88a8 ("power: sequencing: request the WLAN enable GPIO
as-is") broke WLAN on boards on which the wlan-enable GPIO enabling the
wifi module isn't in output mode by default. We need to set direction to
output while retaining the value that was already set to keep the ath
module on if it's already started.

Fixes: a9aaf1ff88a8 ("power: sequencing: request the WLAN enable GPIO as-is")
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Bartosz Golaszewski <[email protected]>
6 months agoMerge tag 'usb-6.11-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Sat, 31 Aug 2024 19:06:28 +0000 (07:06 +1200)]
Merge tag 'usb-6.11-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb

Pull USB fixes from Greg KH:
 "Here are some small USB fixes for 6.11-rc6.  Included in here are:

   - dwc3 driver fixes for reported issues

   - MAINTAINER file update, marking a driver as unsupported :(

   - cdnsp driver fixes

   - USB gadget driver fix

   - USB sysfs fix

   - other tiny fixes

   - new device ids for usb serial driver

  All of these have been in linux-next this week with no reported
  issues"

* tag 'usb-6.11-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
  USB: serial: option: add MeiG Smart SRM825L
  usb: cdnsp: fix for Link TRB with TC
  usb: dwc3: st: add missing depopulate in probe error path
  usb: dwc3: st: fix probed platform device ref count on probe error path
  usb: dwc3: ep0: Don't reset resource alloc flag (including ep0)
  usb: core: sysfs: Unmerge @usb3_hardware_lpm_attr_group in remove_power_attributes()
  usb: typec: fsa4480: Relax CHIP_ID check
  usb: dwc3: xilinx: add missing depopulate in probe error path
  usb: dwc3: omap: add missing depopulate in probe error path
  dt-bindings: usb: microchip,usb2514: Fix reference USB device schema
  usb: gadget: uvc: queue pump work in uvcg_video_enable()
  cdc-acm: Add DISABLE_ECHO quirk for GE HealthCare UI Controller
  usb: cdnsp: fix incorrect index in cdnsp_get_hw_deq function
  usb: dwc3: core: Prevent USB core invalid event buffer address access
  MAINTAINERS: Mark UVC gadget driver as orphan

6 months agoMerge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Sat, 31 Aug 2024 19:00:38 +0000 (07:00 +1200)]
Merge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "Minor fixes only.

  The sd.c one ignores a sync cache request if format is in progress
  which can happen if formatting a drive across suspend/resume"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: sd: Ignore command SYNCHRONIZE CACHE error if format in progress
  scsi: aacraid: Fix double-free on probe failure
  scsi: lpfc: Fix overflow build issue

6 months agoMerge tag 'nfsd-6.11-3' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux
Linus Torvalds [Sat, 31 Aug 2024 18:55:47 +0000 (06:55 +1200)]
Merge tag 'nfsd-6.11-3' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux

Pull nfsd fix from Chuck Lever:

 - One more write delegation fix

* tag 'nfsd-6.11-3' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux:
  nfsd: fix nfsd4_deleg_getattr_conflict in presence of third party lease

6 months agoMerge tag 'xfs-6.11-fixes-4' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Linus Torvalds [Sat, 31 Aug 2024 18:48:37 +0000 (06:48 +1200)]
Merge tag 'xfs-6.11-fixes-4' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux

Pull xfs fixes from Chandan Babu:

 - Do not call out v1 inodes with non-zero di_nlink field as being
   corrupt

 - Change xfs_finobt_count_blocks() to count "free inode btree" blocks
   rather than "inode btree" blocks

 - Don't report the number of trimmed bytes via FITRIM because the
   underlying storage isn't required to do anything and failed discard
   IOs aren't reported to the caller anyway

 - Fix incorrect setting of rm_owner field in an rmap query

 - Report missing disk offset range in an fsmap query

 - Obtain m_growlock when extending realtime section of the filesystem

 - Reset rootdir extent size hint after extending realtime section of
   the filesystem

* tag 'xfs-6.11-fixes-4' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  xfs: reset rootdir extent size hint after growfsrt
  xfs: take m_growlock when running growfsrt
  xfs: Fix missing interval for missing_owner in xfs fsmap
  xfs: use XFS_BUF_DADDR_NULL for daddrs in getfsmap code
  xfs: Fix the owner setting issue for rmap query in xfs fsmap
  xfs: don't bother reporting blocks trimmed via FITRIM
  xfs: xfs_finobt_count_blocks() walks the wrong btree
  xfs: fix folio dirtying for XFILE_ALLOC callers
  xfs: fix di_onlink checking for V1/V2 inodes

6 months agoMerge tag 'arm-fixes-6.11-2' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Linus Torvalds [Sat, 31 Aug 2024 18:42:13 +0000 (06:42 +1200)]
Merge tag 'arm-fixes-6.11-2' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc

Pull ARM SoC fixes from Arnd Bergmann:
 "There is a fairly large number of bug fixes for Qualcomm platforms,
  most of them addressing issues with the devicetree files for the newly
  added Snapdragon X1 based laptops to make them more reliable.

  The Qualcomm driver changes address a few build-time issues as well as
  runtime problems in the tzmem and scm firmware, the USB Type-C driver,
  and the cmd-db and pmic_glink soc drivers.

  The NXP i.MX usually gets a bunch of devicetree fixes that is
  proportional to the number of supported machines. This includes both
  warning fixes and correctness for the 64-bit i.MX9, i.MX8 and
  layerscape platforms, as well as a single fix for a 32-bit i.MX6 based
  board.

  The other changes are the usual minor changes, including an update to
  the MAINTAINERS file, an omap3 dts file and a SoC driver for mpfs
  (risc-v)"

* tag 'arm-fixes-6.11-2' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc: (50 commits)
  firmware: microchip: fix incorrect error report of programming:timeout on success
  soc: qcom: pd-mapper: Fix singleton refcount
  firmware: qcom: tzmem: disable sdm670 platform
  soc: qcom: pmic_glink: Actually communicate when remote goes down
  usb: typec: ucsi: Move unregister out of atomic section
  soc: qcom: pmic_glink: Fix race during initialization
  firmware: qcom: qseecom: remove unused functions
  firmware: qcom: tzmem: fix virtual-to-physical address conversion
  firmware: qcom: scm: Mark get_wq_ctx() as atomic call
  arm64: dts: qcom: x1e80100: Fix Adreno SMMU global interrupt
  arm64: dts: qcom: disable GPU on x1e80100 by default
  arm64: dts: imx8mm-phygate: fix typo pinctrcl-0
  arm64: dts: imx95: correct L3Cache cache-sets
  arm64: dts: imx95: correct a55 power-domains
  arm64: dts: freescale: imx93-tqma9352-mba93xxla: fix typo
  arm64: dts: freescale: imx93-tqma9352: fix CMA alloc-ranges
  ARM: dts: imx6dl-yapp43: Increase LED current to match the yapp4 HW design
  arm64: dts: imx93: update default value for snps,clk-csr
  arm64: dts: freescale: tqma9352: Fix watchdog reset
  arm64: dts: imx8mp-beacon-kit: Fix Stereo Audio on WM8962
  ...

6 months agoMerge tag 'input-for-v6.11-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 31 Aug 2024 03:32:38 +0000 (15:32 +1200)]
Merge tag 'input-for-v6.11-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input

Pull input fix from Dmitry Torokhov:

 - a fix for Cypress PS/2 touchpad for regression introduced in 6.11
   merge window where a timeout condition is incorrectly reported for
   all extended Cypress commands

* tag 'input-for-v6.11-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: cypress_ps2 - fix waiting for command response

6 months agoMerge tag 'pci-v6.11-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/pci/pci
Linus Torvalds [Sat, 31 Aug 2024 02:54:11 +0000 (14:54 +1200)]
Merge tag 'pci-v6.11-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/pci/pci

Pull pci fixes from Bjorn Helgaas:

 - Add Manivannan Sadhasivam as PCI native host bridge and endpoint
   driver reviewer (Manivannan Sadhasivam)

 - Disable MHI RAM data parity error interrupt for qcom SA8775P SoC to
   work around hardware erratum that causes a constant stream of
   interrupts (Manivannan Sadhasivam)

 - Don't try to fall back to qcom Operating Performance Points (OPP)
   support unless the platform actually supports OPP (Manivannan
   Sadhasivam)

 - Add [email protected] mailing list to MAINTAINERS for NXP
   layerscape and imx6 PCI controller drivers (Frank Li)

* tag 'pci-v6.11-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/pci/pci:
  MAINTAINERS: PCI: Add NXP PCI controller mailing list [email protected]
  PCI: qcom: Use OPP only if the platform supports it
  PCI: qcom-ep: Disable MHI RAM data parity error interrupt for SA8775P SoC
  MAINTAINERS: Add Manivannan Sadhasivam as Reviewer for PCI native host bridge and endpoint drivers

6 months agoMerge tag 'block-6.11-20240830' of git://git.kernel.dk/linux
Linus Torvalds [Sat, 31 Aug 2024 01:54:05 +0000 (13:54 +1200)]
Merge tag 'block-6.11-20240830' of git://git.kernel.dk/linux

Pull block fix from Jens Axboe:
 "Fix for a single regression for WRITE_SAME introduced in the 6.11
  merge window"

* tag 'block-6.11-20240830' of git://git.kernel.dk/linux:
  block: fix detection of unsupported WRITE SAME in blkdev_issue_write_zeroes

6 months agoMerge tag 'io_uring-6.11-20240830' of git://git.kernel.dk/linux
Linus Torvalds [Sat, 31 Aug 2024 01:51:27 +0000 (13:51 +1200)]
Merge tag 'io_uring-6.11-20240830' of git://git.kernel.dk/linux

Pull io_uring fixes from Jens Axboe:

 - A fix for a regression that happened in 6.11 merge window, where the
   copying of iovecs for compat mode applications got broken for certain
   cases.

 - Fix for a bug introduced in 6.10, where if using recv/send bundles
   with classic provided buffers, the recv/send would fail to set the
   right iovec count. This caused 0 byte send/recv results. Found via
   code coverage testing and writing a test case to exercise it.

* tag 'io_uring-6.11-20240830' of git://git.kernel.dk/linux:
  io_uring/kbuf: return correct iovec count from classic buffer peek
  io_uring/rsrc: ensure compat iovecs are copied correctly

6 months agoMerge tag 'at91-fixes-6.11' of https://git.kernel.org/pub/scm/linux/kernel/git/at91...
Arnd Bergmann [Fri, 30 Aug 2024 19:52:28 +0000 (19:52 +0000)]
Merge tag 'at91-fixes-6.11' of https://git.kernel.org/pub/scm/linux/kernel/git/at91/linux into arm/fixes

Microchip AT91 fixes for v6.11

It contains:
- DTS directory update to match all entries not only those starting with
  at91 or sama

6 months agoMerge tag 'lsm-pr-20240830' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/lsm
Linus Torvalds [Fri, 30 Aug 2024 18:33:59 +0000 (06:33 +1200)]
Merge tag 'lsm-pr-20240830' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/lsm

Pull lsm fix from Paul Moore:
 "One small patch to correct a NFS permissions problem with SELinux and
  Smack"

* tag 'lsm-pr-20240830' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/lsm:
  selinux,smack: don't bypass permissions check in inode_setsecctx hook

6 months agoMerge tag 'pm-6.11-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Linus Torvalds [Fri, 30 Aug 2024 18:25:34 +0000 (06:25 +1200)]
Merge tag 'pm-6.11-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull power management fixes from Rafael Wysocki:
 "These fix three issues in the amd-pstate cpufreq driver.

  Specifics:

   - Remove checks for highest performance match on preferred cores when
     updating preferred core ranking in amd-pstate (Mario Limonciello)

   - Make amd-pstate call topology_logical_package_id() instead of
     logical_die_id() to get a socked ID for a CPU (Gautham Shenoy)

   - Fix uninitialized variable in amd_pstate_cpu_boost_update() (Dan
     Carpenter)"

* tag 'pm-6.11-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  cpufreq/amd-pstate-ut: Don't check for highest perf matching on prefcore
  cpufreq/amd-pstate: Use topology_logical_package_id() instead of logical_die_id()
  cpufreq: amd-pstate: Fix uninitialized variable in amd_pstate_cpu_boost_update()

6 months agoMerge tag 'dmaengine-fix-6.11' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul...
Linus Torvalds [Fri, 30 Aug 2024 18:20:48 +0000 (06:20 +1200)]
Merge tag 'dmaengine-fix-6.11' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine

Pull dmaengine fixes from Vinod Koul:

 - A bunch of dw driver changes to fix the src/dst addr width config

 - Omap driver fix for sglen initialization

 - stm32-dma3 driver lli_size init fix

 - dw edma driver fixes for watermark interrupts and unmasking STOP and
   ABORT interrupts

* tag 'dmaengine-fix-6.11' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine:
  dmaengine: dw-edma: Do not enable watermark interrupts for HDMA
  dmaengine: dw-edma: Fix unmasking STOP and ABORT interrupts for HDMA
  dmaengine: stm32-dma3: Set lli_size after allocation
  dmaengine: ti: omap-dma: Initialize sglen after allocation
  dmaengine: dw: Unify ret-val local variables naming
  dmaengine: dw: Simplify max-burst calculation procedure
  dmaengine: dw: Define encode_maxburst() above prepare_ctllo() callbacks
  dmaengine: dw: Simplify prepare CTL_LO methods
  dmaengine: dw: Add memory bus width verification
  dmaengine: dw: Add peripheral bus width verification

6 months agoMerge tag 'phy-fixes-6.11' of git://git.kernel.org/pub/scm/linux/kernel/git/phy/linux-phy
Linus Torvalds [Fri, 30 Aug 2024 18:18:07 +0000 (06:18 +1200)]
Merge tag 'phy-fixes-6.11' of git://git.kernel.org/pub/scm/linux/kernel/git/phy/linux-phy

Pull phy fixes from Vinod Koul:

 - Qualcomm QMP X1E80100 PCIe Gen4 PHY initialisation fix

 - Freescale imx8mq tuning parameter name fix

 - Samsung exynos5 fir for error code in probe()

 - Xilinx Zynqmp SGMII linkup failure fix

* tag 'phy-fixes-6.11' of git://git.kernel.org/pub/scm/linux/kernel/git/phy/linux-phy:
  phy: xilinx: phy-zynqmp: Fix SGMII linkup failure on resume
  phy: exynos5-usbdrd: fix error code in probe()
  phy: fsl-imx8mq-usb: fix tuning parameter name
  phy: qcom: qmp-pcie: Fix X1E80100 PCIe Gen4 PHY initialisation

6 months agoMerge tag 'soundwire-6.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 30 Aug 2024 18:15:02 +0000 (06:15 +1200)]
Merge tag 'soundwire-6.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/soundwire

Pull soundwire fix from Vinod Koul:

 - Single fix for non-continous port map programming

* tag 'soundwire-6.11-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/soundwire:
  soundwire: stream: fix programming slave ports for non-continous port maps

6 months agoMerge tag 'iommu-fixes-v6.11-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 30 Aug 2024 18:11:34 +0000 (06:11 +1200)]
Merge tag 'iommu-fixes-v6.11-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/iommu/linux

Pull iommu fixes from Joerg Roedel:

 - Fix a device-stall problem in bad io-page-fault setups (faults
   received from devices with no supporting domain attached).

 - Context flush fix for Intel VT-d.

 - Do not allow non-read+non-write mapping through iommufd as most
   implementations can not handle that.

 - Fix a possible infinite-loop issue in map_pages() path.

 - Add Jean-Philippe as reviewer for SMMUv3 SVA support

* tag 'iommu-fixes-v6.11-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/iommu/linux:
  MAINTAINERS: Add Jean-Philippe as SMMUv3 SVA reviewer
  iommu: Do not return 0 from map_pages if it doesn't do anything
  iommufd: Do not allow creating areas without READ or WRITE
  iommu/vt-d: Fix incorrect domain ID in context flush helper
  iommu: Handle iommu faults for a bad iopf setup

6 months agoMAINTAINERS: PCI: Add NXP PCI controller mailing list [email protected]
Frank Li [Mon, 26 Aug 2024 20:27:39 +0000 (16:27 -0400)]
MAINTAINERS: PCI: Add NXP PCI controller mailing list [email protected]

Add imx mailing list [email protected] for PCI controller of NXP chips
(Layerscape and iMX).

Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Frank Li <[email protected]>
Signed-off-by: Bjorn Helgaas <[email protected]>
Acked-by: Richard Zhu <[email protected]>
6 months agoio_uring/kbuf: return correct iovec count from classic buffer peek
Jens Axboe [Fri, 30 Aug 2024 16:45:54 +0000 (10:45 -0600)]
io_uring/kbuf: return correct iovec count from classic buffer peek

io_provided_buffers_select() returns 0 to indicate success, but it should
be returning 1 to indicate that 1 vec was mapped. This causes peeking
to fail with classic provided buffers, and while that's not a use case
that anyone should use, it should still work correctly.

The end result is that no buffer will be selected, and hence a completion
with '0' as the result will be posted, without a buffer attached.

Fixes: 35c8711c8fc4 ("io_uring/kbuf: add helpers for getting/peeking multiple buffers")
Signed-off-by: Jens Axboe <[email protected]>
6 months agonfsd: fix nfsd4_deleg_getattr_conflict in presence of third party lease
NeilBrown [Wed, 28 Aug 2024 23:06:28 +0000 (09:06 +1000)]
nfsd: fix nfsd4_deleg_getattr_conflict in presence of third party lease

It is not safe to dereference fl->c.flc_owner without first confirming
fl->fl_lmops is the expected manager.  nfsd4_deleg_getattr_conflict()
tests fl_lmops but largely ignores the result and assumes that flc_owner
is an nfs4_delegation anyway.  This is wrong.

With this patch we restore the "!= &nfsd_lease_mng_ops" case to behave
as it did before the change mentioned below.  This is the same as the
current code, but without any reference to a possible delegation.

Fixes: c5967721e106 ("NFSD: handle GETATTR conflict with write delegation")
Signed-off-by: NeilBrown <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
Signed-off-by: Chuck Lever <[email protected]>
6 months agoregulator: core: Stub devm_regulator_bulk_get_const() if !CONFIG_REGULATOR
Douglas Anderson [Fri, 30 Aug 2024 14:35:12 +0000 (07:35 -0700)]
regulator: core: Stub devm_regulator_bulk_get_const() if !CONFIG_REGULATOR

When adding devm_regulator_bulk_get_const() I missed adding a stub for
when CONFIG_REGULATOR is not enabled. Under certain conditions (like
randconfig testing) this can cause the compiler to reports errors
like:

  error: implicit declaration of function 'devm_regulator_bulk_get_const';
  did you mean 'devm_regulator_bulk_get_enable'?

Add the stub.

Fixes: 1de452a0edda ("regulator: core: Allow drivers to define their init data as const")
Reported-by: kernel test robot <[email protected]>
Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
Cc: Neil Armstrong <[email protected]>
Signed-off-by: Douglas Anderson <[email protected]>
Link: https://patch.msgid.link/20240830073511.1.Ib733229a8a19fad8179213c05e1af01b51e42328@changeid
Signed-off-by: Mark Brown <[email protected]>
6 months agoio_uring/rsrc: ensure compat iovecs are copied correctly
Jens Axboe [Wed, 28 Aug 2024 15:42:33 +0000 (09:42 -0600)]
io_uring/rsrc: ensure compat iovecs are copied correctly

For buffer registration (or updates), a userspace iovec is copied in
and updated. If the application is within a compat syscall, then the
iovec type is compat_iovec rather than iovec. However, the type used
in __io_sqe_buffers_update() and io_sqe_buffers_register() is always
struct iovec, and hence the source is incremented by the size of a
non-compat iovec in the loop. This misses every other iovec in the
source, and will run into garbage half way through the copies and
return -EFAULT to the application.

Maintain the source address separately and assign to our user vec
pointer, so that copies always happen from the right source address.

While in there, correct a bad placement of __user which triggered
the following sparse warning prior to this fix:

io_uring/rsrc.c:981:33: warning: cast removes address space '__user' of expression
io_uring/rsrc.c:981:30: warning: incorrect type in assignment (different address spaces)
io_uring/rsrc.c:981:30:    expected struct iovec const [noderef] __user *uvec
io_uring/rsrc.c:981:30:    got struct iovec *[noderef] __user

Fixes: f4eaf8eda89e ("io_uring/rsrc: Drop io_copy_iov in favor of iovec API")
Reviewed-by: Gabriel Krisman Bertazi <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
6 months agoMerge tag 'usb-serial-6.11-rc6' of ssh://gitolite.kernel.org/pub/scm/linux/kernel...
Greg Kroah-Hartman [Fri, 30 Aug 2024 13:41:18 +0000 (15:41 +0200)]
Merge tag 'usb-serial-6.11-rc6' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/johan/usb-serial into usb-linus

Johan writes:

USB-serial device id for 6.11-rc6

Here's a new modem device id.

This one has been in linux-next with no reported issues.

* tag 'usb-serial-6.11-rc6' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/johan/usb-serial:
  USB: serial: option: add MeiG Smart SRM825L

6 months agomm: Fix filemap_invalidate_inode() to use invalidate_inode_pages2_range()
David Howells [Wed, 28 Aug 2024 21:02:45 +0000 (22:02 +0100)]
mm: Fix filemap_invalidate_inode() to use invalidate_inode_pages2_range()

Fix filemap_invalidate_inode() to use invalidate_inode_pages2_range()
rather than truncate_inode_pages_range().  The latter clears the
invalidated bit of a partial pages rather than discarding it entirely.
This causes copy_file_range() to fail on cifs because the partial pages at
either end of the destination range aren't evicted and reread, but rather
just partly cleared.

This causes generic/075 and generic/112 xfstests to fail.

Fixes: 74e797d79cf1 ("mm: Provide a means of invalidation without using launder_folio")
Signed-off-by: David Howells <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
cc: Matthew Wilcox <[email protected]>
cc: Miklos Szeredi <[email protected]>
cc: Trond Myklebust <[email protected]>
cc: Christoph Hellwig <[email protected]>
cc: Andrew Morton <[email protected]>
cc: Alexander Viro <[email protected]>
cc: Christian Brauner <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
Signed-off-by: Christian Brauner <[email protected]>
6 months agoplatform/x86: dell-smbios: Fix error path in dell_smbios_init()
Aleksandr Mishin [Fri, 30 Aug 2024 06:54:28 +0000 (09:54 +0300)]
platform/x86: dell-smbios: Fix error path in dell_smbios_init()

In case of error in build_tokens_sysfs(), all the memory that has been
allocated is freed at end of this function. But then free_group() is
called which performs memory deallocation again.

Also, instead of free_group() call, there should be exit_dell_smbios_smm()
and exit_dell_smbios_wmi() calls, since there is initialization, but there
is no release of resources in case of an error.

Fix these issues by replacing free_group() call with
exit_dell_smbios_wmi() and exit_dell_smbios_smm().

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Fixes: 33b9ca1e53b4 ("platform/x86: dell-smbios: Add a sysfs interface for SMBIOS tokens")
Signed-off-by: Aleksandr Mishin <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Reviewed-by: Ilpo Järvinen <[email protected]>
Signed-off-by: Ilpo Järvinen <[email protected]>
6 months agoMerge tag 'drm-fixes-2024-08-30' of https://gitlab.freedesktop.org/drm/kernel
Linus Torvalds [Fri, 30 Aug 2024 02:17:30 +0000 (14:17 +1200)]
Merge tag 'drm-fixes-2024-08-30' of https://gitlab.freedesktop.org/drm/kernel

Pull drm fixes from Dave Airlie:
 "Another week, another set of GPU fixes. amdgpu and vmwgfx leading the
  charge, then i915 and xe changes along with v3d and some other bits.
  The TTM revert is due to some stuttering graphical apps probably due
  to longer stalls while prefaulting.

  Seems pretty much where I'd expect things,

  ttm:
   - revert prefault change, caused stutters

  aperture:
   - handle non-VGA devices bettter

  amdgpu:
   - SWSMU gaming stability fix
   - SMU 13.0.7 fix
   - SWSMU documentation alignment fix
   - SMU 14.0.x fixes
   - GC 12.x fix
   - Display fix
   - IP discovery fix
   - SMU 13.0.6 fix

  i915:
   - Fix #11195: The external display connect via USB type-C dock stays
     blank after re-connect the dock
   - Make DSI backlight work for 2G version of Lenovo Yoga Tab 3 X90F
   - Move ARL GuC firmware to correct version

  xe:
   - Invalidate media_gt TLBs
   - Fix HWMON i1 power setup write command

  vmwgfx:
   - prevent unmapping active read buffers
   - fix prime with external buffers
   - disable coherent dumb buffers without 3d

  v3d:
   - disable preemption while updating GPU stats"

* tag 'drm-fixes-2024-08-30' of https://gitlab.freedesktop.org/drm/kernel:
  drm/xe/hwmon: Fix WRITE_I1 param from u32 to u16
  drm/v3d: Disable preemption while updating GPU stats
  drm/amd/pm: Drop unsupported features on smu v14_0_2
  drm/amd/pm: Add support for new P2S table revision
  drm/amdgpu: support for gc_info table v1.3
  drm/amd/display: avoid using null object of framebuffer
  drm/amdgpu/gfx12: set UNORD_DISPATCH in compute MQDs
  drm/amd/pm: update message interface for smu v14.0.2/3
  drm/amdgpu/swsmu: always force a state reprogram on init
  drm/amdgpu/smu13.0.7: print index for profiles
  drm/amdgpu: align pp_power_profile_mode with kernel docs
  drm/i915/dp_mst: Fix MST state after a sink reset
  drm/xe: Invalidate media_gt TLBs
  drm/i915: ARL requires a newer GSC firmware
  drm/i915/dsi: Make Lenovo Yoga Tab 3 X90F DMI match less strict
  video/aperture: optionally match the device in sysfb_disable()
  drm/vmwgfx: Disable coherent dumb buffers without 3d
  drm/vmwgfx: Fix prime with external buffers
  drm/vmwgfx: Prevent unmapping active read buffers
  Revert "drm/ttm: increase ttm pre-fault value to PMD size"

6 months agoksmbd: Unlock on in ksmbd_tcp_set_interfaces()
Dan Carpenter [Thu, 29 Aug 2024 19:22:35 +0000 (22:22 +0300)]
ksmbd: Unlock on in ksmbd_tcp_set_interfaces()

Unlock before returning an error code if this allocation fails.

Fixes: 0626e6641f6b ("cifsd: add server handler for central processing and tranport layers")
Cc: [email protected] # v5.15+
Signed-off-by: Dan Carpenter <[email protected]>
Acked-by: Namjae Jeon <[email protected]>
Signed-off-by: Steve French <[email protected]>
6 months agoksmbd: unset the binding mark of a reused connection
Namjae Jeon [Tue, 27 Aug 2024 12:44:41 +0000 (21:44 +0900)]
ksmbd: unset the binding mark of a reused connection

Steve French reported null pointer dereference error from sha256 lib.
cifs.ko can send session setup requests on reused connection.
If reused connection is used for binding session, conn->binding can
still remain true and generate_preauth_hash() will not set
sess->Preauth_HashValue and it will be NULL.
It is used as a material to create an encryption key in
ksmbd_gen_smb311_encryptionkey. ->Preauth_HashValue cause null pointer
dereference error from crypto_shash_update().

BUG: kernel NULL pointer dereference, address: 0000000000000000
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 0 P4D 0
Oops: 0000 [#1] PREEMPT SMP PTI
CPU: 8 PID: 429254 Comm: kworker/8:39
Hardware name: LENOVO 20MAS08500/20MAS08500, BIOS N2CET69W (1.52 )
Workqueue: ksmbd-io handle_ksmbd_work [ksmbd]
RIP: 0010:lib_sha256_base_do_update.isra.0+0x11e/0x1d0 [sha256_ssse3]
<TASK>
? show_regs+0x6d/0x80
? __die+0x24/0x80
? page_fault_oops+0x99/0x1b0
? do_user_addr_fault+0x2ee/0x6b0
? exc_page_fault+0x83/0x1b0
? asm_exc_page_fault+0x27/0x30
? __pfx_sha256_transform_rorx+0x10/0x10 [sha256_ssse3]
? lib_sha256_base_do_update.isra.0+0x11e/0x1d0 [sha256_ssse3]
? __pfx_sha256_transform_rorx+0x10/0x10 [sha256_ssse3]
? __pfx_sha256_transform_rorx+0x10/0x10 [sha256_ssse3]
_sha256_update+0x77/0xa0 [sha256_ssse3]
sha256_avx2_update+0x15/0x30 [sha256_ssse3]
crypto_shash_update+0x1e/0x40
hmac_update+0x12/0x20
crypto_shash_update+0x1e/0x40
generate_key+0x234/0x380 [ksmbd]
generate_smb3encryptionkey+0x40/0x1c0 [ksmbd]
ksmbd_gen_smb311_encryptionkey+0x72/0xa0 [ksmbd]
ntlm_authenticate.isra.0+0x423/0x5d0 [ksmbd]
smb2_sess_setup+0x952/0xaa0 [ksmbd]
__process_request+0xa3/0x1d0 [ksmbd]
__handle_ksmbd_work+0x1c4/0x2f0 [ksmbd]
handle_ksmbd_work+0x2d/0xa0 [ksmbd]
process_one_work+0x16c/0x350
worker_thread+0x306/0x440
? __pfx_worker_thread+0x10/0x10
kthread+0xef/0x120
? __pfx_kthread+0x10/0x10
ret_from_fork+0x44/0x70
? __pfx_kthread+0x10/0x10
ret_from_fork_asm+0x1b/0x30
</TASK>

Fixes: f5a544e3bab7 ("ksmbd: add support for SMB3 multichannel")
Cc: [email protected] # v5.15+
Signed-off-by: Namjae Jeon <[email protected]>
Signed-off-by: Steve French <[email protected]>
6 months agosmb: Annotate struct xattr_smb_acl with __counted_by()
Thorsten Blum [Fri, 23 Aug 2024 11:47:05 +0000 (13:47 +0200)]
smb: Annotate struct xattr_smb_acl with __counted_by()

Add the __counted_by compiler attribute to the flexible array member
entries to improve access bounds-checking via CONFIG_UBSAN_BOUNDS and
CONFIG_FORTIFY_SOURCE.

Signed-off-by: Thorsten Blum <[email protected]>
Acked-by: Namjae Jeon <[email protected]>
Signed-off-by: Steve French <[email protected]>
6 months agoMerge tag 'drm-misc-fixes-2024-08-29' of https://gitlab.freedesktop.org/drm/misc...
Dave Airlie [Fri, 30 Aug 2024 01:28:00 +0000 (11:28 +1000)]
Merge tag 'drm-misc-fixes-2024-08-29' of https://gitlab.freedesktop.org/drm/misc/kernel into drm-fixes

A revert for a previous TTM commit causing stuttering, 3 fixes for
vmwgfx related to buffer operations, a fix for video/aperture with
non-VGA primary devices, and a preemption status fix for v3d

Signed-off-by: Dave Airlie <[email protected]>
From: Maxime Ripard <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/20240829-efficient-swift-from-lemuria-f60c05@houat
6 months agoMerge tag 'drm-xe-fixes-2024-08-29' of https://gitlab.freedesktop.org/drm/xe/kernel...
Dave Airlie [Fri, 30 Aug 2024 01:07:55 +0000 (11:07 +1000)]
Merge tag 'drm-xe-fixes-2024-08-29' of https://gitlab.freedesktop.org/drm/xe/kernel into drm-fixes

- Invalidate media_gt TLBs (Brost)
- Fix HWMON i1 power setup write command (Karthik)

Signed-off-by: Dave Airlie <[email protected]>
From: Rodrigo Vivi <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
6 months agoMerge tag 'execve-v6.11-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/kees...
Linus Torvalds [Fri, 30 Aug 2024 00:32:53 +0000 (12:32 +1200)]
Merge tag 'execve-v6.11-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux

Pull execve fix from Kees Cook:

 - binfmt_elf_fdpic: fix AUXV size with ELF_HWCAP2 (Max Filippov)

* tag 'execve-v6.11-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux:
  binfmt_elf_fdpic: fix AUXV size calculation when ELF_HWCAP2 is defined

6 months agodcache: keep dentry_hashtable or d_hash_shift even when not used
Stephen Brennan [Thu, 29 Aug 2024 18:20:49 +0000 (11:20 -0700)]
dcache: keep dentry_hashtable or d_hash_shift even when not used

The runtime constant feature removes all the users of these variables,
allowing the compiler to optimize them away.  It's quite difficult to
extract their values from the kernel text, and the memory saved by
removing them is tiny, and it was never the point of this optimization.

Since the dentry_hashtable is a core data structure, it's valuable for
debugging tools to be able to read it easily.  For instance, scripts
built on drgn, like the dentrycache script[1], rely on it to be able to
perform diagnostics on the contents of the dcache.  Annotate it as used,
so the compiler doesn't discard it.

Link: https://github.com/oracle-samples/drgn-tools/blob/3afc56146f54d09dfd1f6d3c1b7436eda7e638be/drgn_tools/dentry.py#L325-L355
Fixes: e3c92e81711d ("runtime constants: add x86 architecture support")
Signed-off-by: Stephen Brennan <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
This page took 0.174341 seconds and 4 git commands to generate.