H.J. Lu [Thu, 20 Jan 2022 02:09:40 +0000 (18:09 -0800)]
fs/binfmt_elf: use PT_LOAD p_align values for static PIE
Extend commit ce81bb256a22 ("fs/binfmt_elf: use PT_LOAD p_align values
for suitable start address") which fixed PIE binaries built with
-Wl,-z,max-page-size=0x200000, to cover static PIE binaries. This
fixes:
Jerome Forissier [Thu, 20 Jan 2022 02:09:31 +0000 (18:09 -0800)]
checkpatch: relax regexp for COMMIT_LOG_LONG_LINE
One exceptions to the COMMIT_LOG_LONG_LINE rule is a file path followed
by ':'. That is typically some sort diagnostic message from a compiler
or a build tool, in which case we don't want to wrap the lines but keep
the message unmodified.
The regular expression used to match this pattern currently doesn't
accept absolute paths or + characters. This can result in false
positives as in the following (out-of-tree) example:
...
/home/jerome/work/optee_repo_qemu/build/../toolchains/aarch32/bin/arm-linux-gnueabihf-ld.bfd: /home/jerome/work/toolchains-gcc10.2/aarch32/bin/../lib/gcc/arm-none-linux-gnueabihf/10.2.1/../../../../arm-none-linux-gnueabihf/lib/libstdc++.a(eh_alloc.o): in function `__cxa_allocate_exception':
/tmp/dgboter/bbs/build03--cen7x86_64/buildbot/cen7x86_64--arm-none-linux-gnueabihf/build/src/gcc/libstdc++-v3/libsupc++/eh_alloc.cc:284: undefined reference to `malloc'
...
Update the regular expression to match the above paths.
Isabella Basso [Thu, 20 Jan 2022 02:09:15 +0000 (18:09 -0800)]
test_hash.c: refactor into kunit
Use KUnit framework to make tests more easily integrable with CIs. Even
though these tests are not yet properly written as unit tests this
change should help in debugging.
Also remove kernel messages (i.e. through pr_info) as KUnit handles all
debugging output and let it handle module init and exit details.
Isabella Basso [Thu, 20 Jan 2022 02:09:09 +0000 (18:09 -0800)]
test_hash.c: split test_hash_init
Split up test_hash_init so that it calls each test more explicitly
insofar it is possible without rewriting the entire file. This aims at
improving readability.
Split tests performed on string_or as they don't interfere with those
performed in hash_or. Also separate pr_info calls about skipped tests
as they're not part of the tests themselves, but only warn about
(un)defined arch-specific hash functions.
Isabella Basso [Thu, 20 Jan 2022 02:09:05 +0000 (18:09 -0800)]
test_hash.c: split test_int_hash into arch-specific functions
Split the test_int_hash function to keep its mainloop separate from
arch-specific chunks, which are only compiled as needed. This aims at
improving readability.
Isabella Basso [Thu, 20 Jan 2022 02:09:02 +0000 (18:09 -0800)]
hash.h: remove unused define directive
Patch series "test_hash.c: refactor into KUnit", v3.
We refactored the lib/test_hash.c file into KUnit as part of the student
group LKCAMP [1] introductory hackathon for kernel development.
This test was pointed to our group by Daniel Latypov [2], so its full
conversion into a pure KUnit test was our goal in this patch series, but
we ran into many problems relating to it not being split as unit tests,
which complicated matters a bit, as the reasoning behind the original
tests is quite cryptic for those unfamiliar with hash implementations.
Some interesting developments we'd like to highlight are:
- In patch 1/5 we noticed that there was an unused define directive
that could be removed.
- In patch 4/5 we noticed how stringhash and hash tests are all under
the lib/test_hash.c file, which might cause some confusion, and we
also broke those kernel config entries up.
Overall KUnit developments have been made in the other patches in this
series:
In patches 2/5, 3/5 and 5/5 we refactored the lib/test_hash.c file so as
to make it more compatible with the KUnit style, whilst preserving the
original idea of the maintainer who designed it (i.e. George Spelvin),
which might be undesirable for unit tests, but we assume it is enough
for a first patch.
This patch (of 5):
Currently, there exist hash_32() and __hash_32() functions, which were
introduced in a patch [1] targeting architecture specific optimizations.
These functions can be overridden on a per-architecture basis to achieve
such optimizations. They must set their corresponding define directive
(HAVE_ARCH_HASH_32 and HAVE_ARCH__HASH_32, respectively) so that header
files can deal with these overrides properly.
As the supported 32-bit architectures that have their own hash function
implementation (i.e. m68k, Microblaze, H8/300, pa-risc) have only been
making use of the (more general) __hash_32() function (which only lacks
a right shift operation when compared to the hash_32() function), remove
the define directive corresponding to the arch-specific hash_32()
implementation.
Zhen Lei [Thu, 20 Jan 2022 02:08:59 +0000 (18:08 -0800)]
lib/list_debug.c: print more list debugging context in __list_del_entry_valid()
Currently, the entry->prev and entry->next are considered to be valid as
long as they are not LIST_POISON{1|2}. However, the memory may be
corrupted. The prev->next is invalid probably because 'prev' is
invalid, not because prev->next's content is illegal.
Unfortunately, the printk and its subfunctions will modify the registers
that hold the 'prev' and 'next', and we don't see this valuable
information in the BUG context.
So print the contents of 'entry->prev' and 'entry->next'.
At first, I thought prev->next was overwritten. Later, I carefully
analyzed the RCU code and the disassembly code. The error occurred when
deleting a node from the list rcu_state.gp_wq. The System.map shows
that the address of rcu_state is c0840c00. Then I use gdb to obtain the
offset of rcu_state.gp_wq.task_list.
Because rcu_state.gp_wq has at most one node, so I can guess that "prev
= &rcu_state.gp_wq.task_list". But for other scenes, maybe I wasn't so
lucky, I cannot figure out the value of 'prev'.
Randy Dunlap [Thu, 20 Jan 2022 02:08:50 +0000 (18:08 -0800)]
get_maintainer: don't remind about no git repo when --nogit is used
When --nogit is used with scripts/get_maintainer.pl, the script spews 4
lines of unnecessary information (noise). Do not print those lines when
--nogit is specified.
This change removes the printing of these 4 lines:
./scripts/get_maintainer.pl: No supported VCS found. Add --nogit to options?
Using a git repository produces better results.
Try Linus Torvalds' latest git repository using:
git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
Davidlohr Bueso [Thu, 20 Jan 2022 02:08:47 +0000 (18:08 -0800)]
kernel/sys.c: only take tasklist_lock for get/setpriority(PRIO_PGRP)
PRIO_PGRP needs the tasklist_lock mainly to serialize vs setpgid(2), to
protect against any concurrent change_pid(PIDTYPE_PGID) that can move
the task from one hlist to another while iterating.
However, the remaining can only rely only on RCU:
PRIO_PROCESS only does the task lookup and never iterates over tasklist
and we already have an rcu-aware stable pointer.
PRIO_USER is already racy vs setuid(2) so with creds being rcu
protected, we can end up seeing stale data. When removing the
tasklist_lock there can be a race with (i) fork but this is benign as
the child's nice is inherited and the new task is not observable by the
user yet either, hence the return semantics do not differ. And (ii) a
race with exit, which is a small window and can cause us to miss a task
which was removed from the list and it had the highest nice.
Similarly change the buggy do_each_thread/while_each_thread combo in
PRIO_USER for the rcu-safe for_each_process_thread flavor, which doesn't
make use of next_thread/p->thread_group.
Yafang Shao [Thu, 20 Jan 2022 02:08:43 +0000 (18:08 -0800)]
kthread: dynamically allocate memory to store kthread's full name
When I was implementing a new per-cpu kthread cfs_migration, I found the
comm of it "cfs_migration/%u" is truncated due to the limitation of
TASK_COMM_LEN. For example, the comm of the percpu thread on CPU10~19
all have the same name "cfs_migration/1", which will confuse the user.
This issue is not critical, because we can get the corresponding CPU
from the task's Cpus_allowed. But for kthreads corresponding to other
hardware devices, it is not easy to get the detailed device info from
task comm, for example,
We can shorten these names to work around this problem, but it may be
not applied to all of the truncated kthreads. Take 'jbd2/nvme0n1p2-'
for example, it is a nice name, and it is not a good idea to shorten it.
One possible way to fix this issue is extending the task comm size, but
as task->comm is used in lots of places, that may cause some potential
buffer overflows. Another more conservative approach is introducing a
new pointer to store kthread's full name if it is truncated, which won't
introduce too much overhead as it is in the non-critical path. Finally
we make a dicision to use the second approach. See also the discussions
in this thread:
https://lore.kernel.org/lkml/20211101060419[email protected]/
After this change, the full name of these truncated kthreads will be
displayed via /proc/[pid]/comm:
Yafang Shao [Thu, 20 Jan 2022 02:08:40 +0000 (18:08 -0800)]
tools/testing/selftests/bpf: replace open-coded 16 with TASK_COMM_LEN
As the sched:sched_switch tracepoint args are derived from the kernel,
we'd better make it same with the kernel. So the macro TASK_COMM_LEN is
converted to type enum, then all the BPF programs can get it through
BTF.
The BPF program which wants to use TASK_COMM_LEN should include the
header vmlinux.h. Regarding the test_stacktrace_map and
test_tracepoint, as the type defined in linux/bpf.h are also defined in
vmlinux.h, so we don't need to include linux/bpf.h again.
Yafang Shao [Thu, 20 Jan 2022 02:08:33 +0000 (18:08 -0800)]
samples/bpf/test_overhead_kprobe_kern: replace bpf_probe_read_kernel with bpf_probe_read_kernel_str to get task comm
bpf_probe_read_kernel_str() will add a nul terminator to the dst, then
we don't care about if the dst size is big enough. This patch also
replaces the hard-coded 16 with TASK_COMM_LEN to make it grepable.
Yafang Shao [Thu, 20 Jan 2022 02:08:26 +0000 (18:08 -0800)]
drivers/infiniband: replace open-coded string copy with get_task_comm
We'd better use the helper get_task_comm() rather than the open-coded
strlcpy() to get task comm. As the comment above the hard-coded 16, we
can replace it with TASK_COMM_LEN.
Yafang Shao [Thu, 20 Jan 2022 02:08:22 +0000 (18:08 -0800)]
fs/exec: replace strncpy with strscpy_pad in __get_task_comm
If the dest buffer size is smaller than sizeof(tsk->comm), the buffer
will be without null ternimator, that may cause problem. Using
strscpy_pad() instead of strncpy() in __get_task_comm() can make the
string always nul ternimated and zero padded.
Yafang Shao [Thu, 20 Jan 2022 02:08:19 +0000 (18:08 -0800)]
fs/exec: replace strlcpy with strscpy_pad in __set_task_comm
Patch series "task comm cleanups", v2.
This patchset is part of the patchset "extend task comm from 16 to
24"[1]. Now we have different opinion that dynamically allocates memory
to store kthread's long name into a separate pointer, so I decide to
take the useful cleanups apart from the original patchset and send it
separately[2].
These useful cleanups can make the usage around task comm less
error-prone. Furthermore, it will be useful if we want to extend task
comm in the future.
strlcpy() can trigger out-of-bound reads on the source string[1], we'd
better use strscpy() instead. To make it be robust against full tsk->comm
copies that got noticed in other places, we should make sure it's zero
padded.
proc/vmcore: don't fake reading zeroes on surprise vmcore_cb unregistration
In commit cc5f2704c934 ("proc/vmcore: convert oldmem_pfn_is_ram callback
to more generic vmcore callbacks"), we added detection of surprise
vmcore_cb unregistration after the vmcore was already opened. Once
detected, we warn the user and simulate reading zeroes from that point
on when accessing the vmcore.
The basic reason was that unexpected unregistration, for example, by
manually unbinding a driver from a device after opening the vmcore, is
not supported and could result in reading oldmem the vmcore_cb would
have actually prohibited while registered. However, something like that
can similarly be trigger by a user that's really looking for trouble
simply by unbinding the relevant driver before opening the vmcore -- or
by disallowing loading the driver in the first place. So it's actually
of limited help.
Currently, unregistration can only be triggered via virtio-mem when
manually unbinding the driver from the device inside the VM; there is no
way to trigger it from the hypervisor, as hypervisors don't allow for
unplugging virtio-mem devices -- ripping out system RAM from a VM
without coordination with the guest is usually not a good idea.
The important part is that unbinding the driver and unregistering the
vmcore_cb while concurrently reading the vmcore won't crash the system,
and that is handled by the rwsem.
To make the mechanism more future proof, let's remove the "read zero"
part, but leave the warning in place. For example, we could have a
future driver (like virtio-balloon) that will contact the hypervisor to
figure out if we already populated a page for a given PFN.
Hotunplugging such a device and consequently unregistering the vmcore_cb
could be triggered from the hypervisor without harming the system even
while kdump is running. In that case, we don't want to silently end up
with a vmcore that contains wrong data, because the user inside the VM
might be unaware of the hypervisor action and might easily miss the
warning in the log.
Kefeng Wang [Thu, 20 Jan 2022 02:07:53 +0000 (18:07 -0800)]
mm: percpu: add generic pcpu_populate_pte() function
With NEED_PER_CPU_PAGE_FIRST_CHUNK enabled, we need a function to
populate pte, this patch adds a generic pcpu populate pte function,
pcpu_populate_pte(), which is marked __weak and used on most
architectures, but it is overridden on x86, which has its own
implementation.
With the previous patch, we could add a generic pcpu first chunk
allocate and free function to cleanup the duplicated definations on each
architecture.
Kefeng Wang [Thu, 20 Jan 2022 02:07:45 +0000 (18:07 -0800)]
mm: percpu: add pcpu_fc_cpu_to_node_fn_t typedef
Add pcpu_fc_cpu_to_node_fn_t and pass it into pcpu_fc_alloc_fn_t, pcpu
first chunk allocation will call it to alloc memblock on the
corresponding node by it, this is prepare for the next patch.
Kefeng Wang [Thu, 20 Jan 2022 02:07:41 +0000 (18:07 -0800)]
mm: percpu: generalize percpu related config
Patch series "mm: percpu: Cleanup percpu first chunk function".
When supporting page mapping percpu first chunk allocator on arm64, we
found there are lots of duplicated codes in percpu embed/page first chunk
allocator. This patchset is aimed to cleanup them and should no function
change.
The currently supported status about 'embed' and 'page' in Archs shows
below,
The pcpu_fc_alloc_fn_t/pcpu_fc_free_fn_t is killed, we provide generic
pcpu_fc_alloc() and pcpu_fc_free() function, which are called in the
pcpu_embed/page_first_chunk().
1) For pcpu_embed_first_chunk(), pcpu_fc_cpu_to_node_fn_t is needed to be
provided when archs supported NUMA.
2) For pcpu_page_first_chunk(), the pcpu_fc_populate_pte_fn_t is killed too,
a generic pcpu_populate_pte() which marked '__weak' is provided, if you
need a different function to populate pte on the arch(like x86), please
provide its own implementation.
The HAVE_SETUP_PER_CPU_AREA/NEED_PER_CPU_EMBED_FIRST_CHUNK/
NEED_PER_CPU_PAGE_FIRST_CHUNK/USE_PERCPU_NUMA_NODE_ID configs, which have
duplicate definitions on platforms that subscribe it.
Move them into mm, drop these redundant definitions and instead just
select it on applicable platforms.
Steve French [Thu, 20 Jan 2022 04:00:29 +0000 (22:00 -0600)]
smb3: send NTLMSSP version information
For improved debugging it can be helpful to send version information
as other clients do during NTLMSSP negotiation. See protocol document
MS-NLMP section 2.2.1.1
Set the major and minor versions based on the kernel version, and the
BuildNumber based on the internal cifs.ko module version number,
and following the recommendation in the protocol documentation
(MS-NLMP section 2.2.10) we set the NTLMRevisionCurrent field to 15.
Palmer Dabbelt [Thu, 20 Jan 2022 03:23:41 +0000 (19:23 -0800)]
RISC-V: Introduce sv48 support without relocatable kernel
This patchset allows to have a single kernel for sv39 and sv48 without
being relocatable.
The idea comes from Arnd Bergmann who suggested to do the same as x86,
that is mapping the kernel to the end of the address space, which allows
the kernel to be linked at the same address for both sv39 and sv48 and
then does not require to be relocated at runtime.
This implements sv48 support at runtime. The kernel will try to boot
with 4-level page table and will fallback to 3-level if the HW does not
support it. Folding the 4th level into a 3-level page table has almost
no cost at runtime.
Note that kasan region had to be moved to the end of the address space
since its location must be known at compile-time and then be valid for
both sv39 and sv48 (and sv57 that is coming).
* riscv-sv48-v3:
riscv: Explicit comment about user virtual address space size
riscv: Use pgtable_l4_enabled to output mmu_type in cpuinfo
riscv: Implement sv48 support
asm-generic: Prepare for riscv use of pud_alloc_one and pud_free
riscv: Allow to dynamically define VA_BITS
riscv: Introduce functions to switch pt_ops
riscv: Split early kasan mapping to prepare sv48 introduction
riscv: Move KASAN mapping next to the kernel mapping
riscv: Get rid of MAXPHYSMEM configs
Alexandre Ghiti [Mon, 6 Dec 2021 10:46:53 +0000 (11:46 +0100)]
riscv: Explicit comment about user virtual address space size
Define precisely the size of the user accessible virtual space size
for sv32/39/48 mmu types and explain why the whole virtual address
space is split into 2 equal chunks between kernel and user space.
Alexandre Ghiti [Mon, 6 Dec 2021 10:46:52 +0000 (11:46 +0100)]
riscv: Use pgtable_l4_enabled to output mmu_type in cpuinfo
Now that the mmu type is determined at runtime using SATP
characteristic, use the global variable pgtable_l4_enabled to output
mmu type of the processor through /proc/cpuinfo instead of relying on
device tree infos.
Alexandre Ghiti [Mon, 6 Dec 2021 10:46:51 +0000 (11:46 +0100)]
riscv: Implement sv48 support
By adding a new 4th level of page table, give the possibility to 64bit
kernel to address 2^48 bytes of virtual address: in practice, that offers
128TB of virtual address space to userspace and allows up to 64TB of
physical memory.
If the underlying hardware does not support sv48, we will automatically
fallback to a standard 3-level page table by folding the new PUD level into
PGDIR level. In order to detect HW capabilities at runtime, we
use SATP feature that ignores writes with an unsupported mode.
Alexandre Ghiti [Mon, 6 Dec 2021 10:46:50 +0000 (11:46 +0100)]
asm-generic: Prepare for riscv use of pud_alloc_one and pud_free
In the following commits, riscv will almost use the generic versions of
pud_alloc_one and pud_free but an additional check is required since those
functions are only relevant when using at least a 4-level page table, which
will be determined at runtime on riscv.
So move the content of those functions into other functions that riscv
can use without duplicating code.
Alexandre Ghiti [Mon, 6 Dec 2021 10:46:48 +0000 (11:46 +0100)]
riscv: Allow to dynamically define VA_BITS
With 4-level page table folding at runtime, we don't know at compile time
the size of the virtual address space so we must set VA_BITS dynamically
so that sparsemem reserves the right amount of memory for struct pages.
Alexandre Ghiti [Mon, 6 Dec 2021 10:46:47 +0000 (11:46 +0100)]
riscv: Introduce functions to switch pt_ops
This simply gathers the different pt_ops initialization in functions
where a comment was added to explain why the page table operations must
be changed along the boot process.
Alexandre Ghiti [Mon, 6 Dec 2021 10:46:46 +0000 (11:46 +0100)]
riscv: Split early kasan mapping to prepare sv48 introduction
Now that kasan shadow region is next to the kernel, for sv48, this
region won't be aligned on PGDIR_SIZE and then when populating this
region, we'll need to get down to lower levels of the page table. So
instead of reimplementing the page table walk for the early population,
take advantage of the existing functions used for the final population.
Note that kasan swapper initialization must also be split since memblock
is not initialized at this point and as the last PGD is shared with the
kernel, we'd need to allocate a PUD so postpone the kasan final
population after the kernel population is done.
Alexandre Ghiti [Mon, 6 Dec 2021 10:46:45 +0000 (11:46 +0100)]
riscv: Move KASAN mapping next to the kernel mapping
Now that KASAN_SHADOW_OFFSET is defined at compile time as a config,
this value must remain constant whatever the size of the virtual address
space, which is only possible by pushing this region at the end of the
address space next to the kernel mapping.
Alexandre Ghiti [Mon, 17 Jan 2022 09:57:16 +0000 (10:57 +0100)]
riscv: Get rid of MAXPHYSMEM configs
CONFIG_MAXPHYSMEM_* are actually never used, even the nommu defconfigs
selecting the MAXPHYSMEM_2GB had no effects on PAGE_OFFSET since it was
preempted by !MMU case right before.
In addition, the move of the kernel mapping at the end of the address
space broke the use of MAXPHYSMEM_2G with MMU since it defines PAGE_OFFSET
at the same address as the kernel mapping.
Brian Foster [Tue, 18 Jan 2022 19:32:35 +0000 (11:32 -0800)]
xfs: flush inodegc workqueue tasks before cancel
The xfs_inodegc_stop() helper performs a high level flush of pending
work on the percpu queues and then runs a cancel_work_sync() on each
of the percpu work tasks to ensure all work has completed before
returning. While cancel_work_sync() waits for wq tasks to complete,
it does not guarantee work tasks have started. This means that the
_stop() helper can queue and instantly cancel a wq task without
having completed the associated work. This can be observed by
tracepoint inspection of a simple "rm -f <file>; fsfreeze -f <mnt>"
test:
The first few lines show that the inode is removed and need inactive
state set, but the inactivation work has not completed before the
inodegc mechanism stops. The inactivation doesn't actually occur
until the fs is unfrozen and the gc mechanism starts back up. Note
that this test requires fsfreeze to reproduce because xfs_freeze
indirectly invokes xfs_fs_statfs(), which calls xfs_inodegc_flush().
When this occurs, the workqueue try_to_grab_pending() logic first
tries to steal the pending bit, which does not succeed because the
bit has been set by queue_work_on(). Subsequently, it checks for
association of a pool workqueue from the work item under the pool
lock. This association is set at the point a work item is queued and
cleared when dequeued for processing. If the association exists, the
work item is removed from the queue and cancel_work_sync() returns
true. If the pwq association is cleared, the remove attempt assumes
the task is busy and retries (eventually returning false to the
caller after waiting for the work task to complete).
To avoid this race, we can flush each work item explicitly before
cancel. However, since the _queue_all() already schedules each
underlying work item, the workqueue level helpers are sufficient to
achieve the same ordering effect. E.g., the inodegc enabled flag
prevents scheduling any further work in the _stop() case. Use the
drain_workqueue() helper in this particular case to make the intent
a bit more self explanatory.
Jisheng Zhang [Mon, 6 Dec 2021 15:03:53 +0000 (23:03 +0800)]
riscv: mm: init: try best to remove #ifdef CONFIG_XIP_KERNEL usage
Currently, the #ifdef CONFIG_XIP_KERNEL usage can be divided into the
following three types:
The first one is for functions/declarations only used in XIP case.
The second one is for XIP_FIXUP case. Something as below:
|foo_type foo;
|#ifdef CONFIG_XIP_KERNEL
|#define foo (*(foo_type *)XIP_FIXUP(&foo))
|#endif
Usually, it's better to let the foo macro sit with the foo var
together. But if various foos are defined adjacently, we can
save some #ifdef CONFIG_XIP_KERNEL usage by grouping them together.
The third one is for different implementations for XIP, usually, this
is a #ifdef...#else...#endif case.
This patch moves the pt_ops macro to adjacent #ifdef CONFIG_XIP_KERNEL
and group first type usage cases into one.
Jisheng Zhang [Mon, 6 Dec 2021 15:03:52 +0000 (23:03 +0800)]
riscv: mm: init: try IS_ENABLED(CONFIG_XIP_KERNEL) instead of #ifdef
Try our best to replace the conditional compilation using
"#ifdef CONFIG_XIP_KERNEL" with "IS_ENABLED(CONFIG_XIP_KERNEL)", to
simplify the code and to increase compile coverage.
Jisheng Zhang [Mon, 6 Dec 2021 15:03:50 +0000 (23:03 +0800)]
riscv: mm: init: try best to use IS_ENABLED(CONFIG_64BIT) instead of #ifdef
Try our best to replace the conditional compilation using
"#ifdef CONFIG_64BIT" by a check for "IS_ENABLED(CONFIG_64BIT)", to
simplify the code and to increase compile coverage.
Now we can also remove the __maybe_unused used in max_mapped_addr
declaration.
We also remove the BUG_ON check of mapping the last 4K bytes of the
addressable memory since this is always true for every kernel actually.
Warning: Kernel ABI header at 'tools/arch/x86/include/uapi/asm/prctl.h' differs from latest version at 'arch/x86/include/uapi/asm/prctl.h'
diff -u tools/arch/x86/include/uapi/asm/prctl.h arch/x86/include/uapi/asm/prctl.h
David Howells [Tue, 17 Nov 2020 15:56:59 +0000 (15:56 +0000)]
cifs: Support fscache indexing rewrite
Change the cifs filesystem to take account of the changes to fscache's
indexing rewrite and reenable caching in cifs.
The following changes have been made:
(1) The fscache_netfs struct is no more, and there's no need to register
the filesystem as a whole.
(2) The session cookie is now an fscache_volume cookie, allocated with
fscache_acquire_volume(). That takes three parameters: a string
representing the "volume" in the index, a string naming the cache to
use (or NULL) and a u64 that conveys coherency metadata for the
volume.
For cifs, I've made it render the volume name string as:
"cifs,<ipaddress>,<sharename>"
where the sharename has '/' characters replaced with ';'.
This probably needs rethinking a bit as the total name could exceed
the maximum filename component length.
Further, the coherency data is currently just set to 0. It needs
something else doing with it - I wonder if it would suffice simply to
sum the resource_id, vol_create_time and vol_serial_number or maybe
hash them.
(3) The fscache_cookie_def is no more and needed information is passed
directly to fscache_acquire_cookie(). The cache no longer calls back
into the filesystem, but rather metadata changes are indicated at
other times.
fscache_acquire_cookie() is passed the same keying and coherency
information as before.
(4) The functions to set/reset cookies are removed and
fscache_use_cookie() and fscache_unuse_cookie() are used instead.
fscache_use_cookie() is passed a flag to indicate if the cookie is
opened for writing. fscache_unuse_cookie() is passed updates for the
metadata if we changed it (ie. if the file was opened for writing).
These are called when the file is opened or closed.
(5) cifs_setattr_*() are made to call fscache_resize() to change the size
of the cache object.
(6) The functions to read and write data are stubbed out pending a
conversion to use netfslib.
Changes
=======
ver #8:
- Abstract cache invalidation into a helper function.
- Fix some checkpatch warnings[3].
ver #7:
- Removed the accidentally added-back call to get the super cookie in
cifs_root_iget().
- Fixed the right call to cifs_fscache_get_super_cookie() to take account
of the "-o fsc" mount flag.
ver #6:
- Moved the change of gfpflags_allow_blocking() to current_is_kswapd() for
cifs here.
- Fixed one of the error paths in cifs_atomic_open() to jump around the
call to use the cookie.
- Fixed an additional successful return in the middle of cifs_open() to
use the cookie on the way out.
- Only get a volume cookie (and thus inode cookies) when "-o fsc" is
supplied to mount.
ver #5:
- Fixed a couple of bits of cookie handling[2]:
- The cookie should be released in cifs_evict_inode(), not
cifsFileInfo_put_final(). The cookie needs to persist beyond file
closure so that writepages will be able to write to it.
- fscache_use_cookie() needs to be called in cifs_atomic_open() as it is
for cifs_open().
ver #4:
- Fixed the use of sizeof with memset.
- tcon->vol_create_time is __le64 so doesn't need cpu_to_le64().
ver #3:
- Canonicalise the cifs coherency data to make the cache portable.
- Set volume coherency data.
ver #2:
- Use gfpflags_allow_blocking() rather than using flag directly.
- Upgraded to -rc4 to allow for upstream changes[1].
- fscache_acquire_volume() now returns errors.
KVM: SVM: Nullify vcpu_(un)blocking() hooks if AVIC is disabled
Nullify svm_x86_ops.vcpu_(un)blocking if AVIC/APICv is disabled as the
hooks are necessary only to clear the vCPU's IsRunning entry in the
Physical APIC and to update IRTE entries if the VM has a pass-through
device attached.
Opportunistically rename the helpers to clarify their AVIC relationship.
KVM: SVM: Move svm_hardware_setup() and its helpers below svm_x86_ops
Move svm_hardware_setup() below svm_x86_ops so that KVM can modify ops
during setup, e.g. the vcpu_(un)blocking hooks can be nullified if AVIC
is disabled or unsupported.
KVM: SVM: Drop AVIC's intermediate avic_set_running() helper
Drop avic_set_running() in favor of calling avic_vcpu_{load,put}()
directly, and modify the block+put path to use preempt_disable/enable()
instead of get/put_cpu(), as it doesn't actually care about the current
pCPU associated with the vCPU. Opportunistically add lockdep assertions
as being preempted in avic_vcpu_put() would lead to consuming stale data,
even though doing so _in the current code base_ would not be fatal.
Add a much needed comment explaining why svm_vcpu_blocking() needs to
unload the AVIC and update the IRTE _before_ the vCPU starts blocking.
KVM: VMX: Don't do full kick when handling posted interrupt wakeup
When waking vCPUs in the posted interrupt wakeup handling, do exactly
that and no more. There is no need to kick the vCPU as the wakeup
handler just needs to get the vCPU task running, and if it's in the guest
then it's definitely running.
KVM: VMX: Don't do full kick when triggering posted interrupt "fails"
Replace the full "kick" with just the "wake" in the fallback path when
triggering a virtual interrupt via a posted interrupt fails because the
guest is not IN_GUEST_MODE. If the guest transitions into guest mode
between the check and the kick, then it's guaranteed to see the pending
interrupt as KVM syncs the PIR to IRR (and onto GUEST_RVI) after setting
IN_GUEST_MODE. Kicking the guest in this case is nothing more than an
unnecessary VM-Exit (and host IRQ).
Opportunistically update comments to explain the various ordering rules
and barriers at play.
KVM: SVM: Skip AVIC and IRTE updates when loading blocking vCPU
Don't bother updating the Physical APIC table or IRTE when loading a vCPU
that is blocking, i.e. won't be marked IsRun{ning}=1, as the pCPU is
queried if and only if IsRunning is '1'. If the vCPU was migrated, the
new pCPU will be picked up when avic_vcpu_load() is called by
svm_vcpu_unblocking().
KVM: SVM: Use kvm_vcpu_is_blocking() in AVIC load to handle preemption
Use kvm_vcpu_is_blocking() to determine whether or not the vCPU should be
marked running during avic_vcpu_load(). Drop avic_is_running, which
really should have been named "vcpu_is_not_blocking", as it tracked if
the vCPU was blocking, not if it was actually running, e.g. it was set
during svm_create_vcpu() when the vCPU was obviously not running.
This is technically a teeny tiny functional change, as the vCPU will be
marked IsRunning=1 on being reloaded if the vCPU is preempted between
svm_vcpu_blocking() and prepare_to_rcuwait(). But that's a benign change
as the vCPU will be marked IsRunning=0 when KVM voluntarily schedules out
the vCPU.
KVM: SVM: Remove unnecessary APICv/AVIC update in vCPU unblocking path
Remove handling of KVM_REQ_APICV_UPDATE from svm_vcpu_unblocking(), it's
no longer needed as it was made obsolete by commit df7e4827c549 ("KVM:
SVM: call avic_vcpu_load/avic_vcpu_put when enabling/disabling AVIC").
Prior to that commit, the manual check was necessary to ensure the AVIC
stuff was updated by avic_set_running() when a request to enable APICv
became pending while the vCPU was blocking, as the request handling
itself would not do the update. But, as evidenced by the commit, that
logic was flawed and subject to various races.
Now that svm_refresh_apicv_exec_ctrl() does avic_vcpu_load/put() in
response to an APICv status change, drop the manual check in the
unblocking path.
KVM: SVM: Don't bother checking for "running" AVIC when kicking for IPIs
Drop the avic_vcpu_is_running() check when waking vCPUs in response to a
VM-Exit due to incomplete IPI delivery. The check isn't wrong per se, but
it's not 100% accurate in the sense that it doesn't guarantee that the vCPU
was one of the vCPUs that didn't receive the IPI.
The check isn't required for correctness as blocking == !running in this
context.
From a performance perspective, waking a live task is not expensive as the
only moderately costly operation is a locked operation to temporarily
disable preemption. And if that is indeed a performance issue,
kvm_vcpu_is_blocking() would be a better check than poking into the AVIC.
KVM: SVM: Signal AVIC doorbell iff vCPU is in guest mode
Signal the AVIC doorbell iff the vCPU is running in the guest. If the vCPU
is not IN_GUEST_MODE, it's guaranteed to pick up any pending IRQs on the
next VMRUN, which unconditionally processes the vIRR.
KVM: VMX: Move preemption timer <=> hrtimer dance to common x86
Handle the switch to/from the hypervisor/software timer when a vCPU is
blocking in common x86 instead of in VMX. Even though VMX is the only
user of a hypervisor timer, the logic and all functions involved are
generic x86 (unless future CPUs do something completely different and
implement a hypervisor timer that runs regardless of mode).
Handling the switch in common x86 will allow for the elimination of the
pre/post_blocks hooks, and also lets KVM switch back to the hypervisor
timer if and only if it was in use (without additional params). Add a
comment explaining why the switch cannot be deferred to kvm_sched_out()
or kvm_vcpu_block().
KVM: Move x86 VMX's posted interrupt list_head to vcpu_vmx
Move the seemingly generic block_vcpu_list from kvm_vcpu to vcpu_vmx, and
rename the list and all associated variables to clarify that it tracks
the set of vCPU that need to be poked on a posted interrupt to the wakeup
vector. The list is not used to track _all_ vCPUs that are blocking, and
the term "blocked" can be misleading as it may refer to a blocking
condition in the host or the guest, where as the PI wakeup case is
specifically for the vCPUs that are actively blocking from within the
guest.
KVM: VMX: Handle PI descriptor updates during vcpu_put/load
Move the posted interrupt pre/post_block logic into vcpu_put/load
respectively, using the kvm_vcpu_is_blocking() to determining whether or
not the wakeup handler needs to be set (and unset). This avoids updating
the PI descriptor if halt-polling is successful, reduces the number of
touchpoints for updating the descriptor, and eliminates the confusing
behavior of intentionally leaving a "stale" PI.NDST when a blocking vCPU
is scheduled back in after preemption.
The downside is that KVM will do the PID update twice if the vCPU is
preempted after prepare_to_rcuwait() but before schedule(), but that's a
rare case (and non-existent on !PREEMPT kernels).
The notable wart is the need to send a self-IPI on the wakeup vector if
an outstanding notification is pending after configuring the wakeup
vector. Ideally, KVM would just do a kvm_vcpu_wake_up() in this case,
but the scheduler doesn't support waking a task from its preemption
notifier callback, i.e. while the task is right in the middle of
being scheduled out.
Note, setting the wakeup vector before halt-polling is not necessary:
once the pending IRQ will be recorded in the PIR, kvm_vcpu_has_events()
will detect this (via kvm_cpu_get_interrupt(), kvm_apic_get_interrupt(),
apic_has_interrupt_for_ppr() and finally vmx_sync_pir_to_irr()) and
terminate the polling.
KVM: selftests: Add a test to force emulation with a pending exception
Add a VMX specific test to verify that KVM doesn't explode if userspace
attempts KVM_RUN when emulation is required with a pending exception.
KVM VMX's emulation support for !unrestricted_guest punts exceptions to
userspace instead of attempting to synthesize the exception with all the
correct state (and stack switching, etc...).
Punting is acceptable as there's never been a request to support
injecting exceptions when emulating due to invalid state, but KVM has
historically assumed that userspace will do the right thing and either
clear the exception or kill the guest. Deliberately do the opposite and
attempt to re-enter the guest with a pending exception and emulation
required to verify KVM continues to punt the combination to userspace,
e.g. doesn't explode, WARN, etc...
KVM: VMX: Reject KVM_RUN if emulation is required with pending exception
Reject KVM_RUN if emulation is required (because VMX is running without
unrestricted guest) and an exception is pending, as KVM doesn't support
emulating exceptions except when emulating real mode via vm86. The vCPU
is hosed either way, but letting KVM_RUN proceed triggers a WARN due to
the impossible condition. Alternatively, the WARN could be removed, but
then userspace and/or KVM bugs would result in the vCPU silently running
in a bad state, which isn't very friendly to users.
Originally, the bug was hit by syzkaller with a nested guest as that
doesn't require kvm_intel.unrestricted_guest=0. That particular flavor
is likely fixed by commit cd0e615c49e5 ("KVM: nVMX: Synthesize
TRIPLE_FAULT for L2 if emulation is required"), but it's trivial to
trigger the WARN with a non-nested guest, and userspace can likely force
bad state via ioctls() for a nested guest as well.
Checking for the impossible condition needs to be deferred until KVM_RUN
because KVM can't force specific ordering between ioctls. E.g. clearing
exception.pending in KVM_SET_SREGS doesn't prevent userspace from setting
it in KVM_SET_VCPU_EVENTS, and disallowing KVM_SET_VCPU_EVENTS with
emulation_required would prevent userspace from queuing an exception and
then stuffing sregs. Note, if KVM were to try and detect/prevent the
condition prior to KVM_RUN, handle_invalid_guest_state() and/or
handle_emulation_failure() would need to be modified to clear the pending
exception prior to exiting to userspace.
Jim Mattson [Sat, 15 Jan 2022 05:24:31 +0000 (21:24 -0800)]
selftests: kvm/x86: Add test for KVM_SET_PMU_EVENT_FILTER
Verify that the PMU event filter works as expected.
Note that the virtual PMU doesn't work as expected on AMD Zen CPUs (an
intercepted rdmsr is counted as a retired branch instruction), but the
PMU event filter does work.
Shyam Prasad N [Mon, 17 Jan 2022 07:15:02 +0000 (07:15 +0000)]
cifs: cifs_ses_mark_for_reconnect should also update reconnect bits
Recent restructuring of cifs_reconnect introduced a helper func
named cifs_ses_mark_for_reconnect, which updates the state of tcp
session for all the channels of a session for reconnect.
However, this does not update the session state and chans_need_reconnect
bitmask. This change fixes that.
Also, cifs_mark_tcp_sess_for_reconnect should mark set the bitmask
for all channels when the whole session is marked for reconnect.
Fixed that here too.
Shyam Prasad N [Tue, 18 Jan 2022 09:24:08 +0000 (09:24 +0000)]
cifs: update tcpStatus during negotiate and sess setup
Till the end of SMB session setup, update tcpStatus and
avoid updating session status field. There was a typo in
cifs_setup_session, which caused ses->status to be updated
instead. This was causing issues during reconnect.
Shyam Prasad N [Sun, 16 Jan 2022 13:38:14 +0000 (13:38 +0000)]
cifs: make status checks in version independent callers
The status of tcp session, smb session and tcon have the
same flow, irrespective of the SMB version used. Hence
these status checks and updates should happen in the
version independent callers of these commands.
Shyam Prasad N [Sun, 16 Jan 2022 13:28:34 +0000 (13:28 +0000)]
cifs: remove repeated state change in dfs tree connect
cifs_tree_connect checks and sets the tidStatus for the tcon.
cifs_tree_connect also calls a dfs specific tree connect
function, which also does similar checks. This should
not happen. Removing it with this change.
Shyam Prasad N [Sun, 16 Jan 2022 11:19:36 +0000 (11:19 +0000)]
cifs: fix the cifs_reconnect path for DFS
Recently, the cifs_reconnect code was refactored into
two branches for regular vs dfs codepath. Some of my
recent changes were missing in the dfs path, namely the
code to enable periodic DNS query, and a missing lock.
Shyam Prasad N [Wed, 17 Nov 2021 15:57:22 +0000 (15:57 +0000)]
cifs: fix the connection state transitions with multichannel
Recent changes to multichannel required some adjustments in
the way connection states transitioned during/after reconnect.
Also some minor fixes:
1. A pending switch of GlobalMid_Lock to cifs_tcp_ses_lock
2. Relocations of the code that logs reconnect
3. Changed some code in allocate_mid to suit the new scheme
projects / linux.git / log