Ravi Bangoria [Thu, 13 Feb 2020 06:43:04 +0000 (12:13 +0530)]
perf annotate: Prefer cmdline option over default config
For all the perf-config options that can also be set from command line
option, the preference is given to command line version in case of any
conflict. But that's opposite in case of perf annotate. i.e. the more
preference is given to default option rather than command line option.
Fix it.
Note that we have removed show_nr_samples and show_total_period from
annotation_options because they are not used. Instead of them we use
symbol_conf.show_nr_samples and symbol_conf.show_total_period.
Committer testing:
Using 'perf annotate --stdio2' to use the TUI rendering but emitting the output to stdio:
Ravi Bangoria [Thu, 13 Feb 2020 06:43:01 +0000 (12:13 +0530)]
perf annotate: Fix --show-nr-samples for tui/stdio2
perf annotate --show-nr-samples does not really show number of samples.
The reason is we have two separate variables for the same purpose.
One is in symbol_conf.show_nr_samples and another is
annotation_options.show_nr_samples.
We save command line option in symbol_conf.show_nr_samples but uses
annotation_option.show_nr_samples while rendering tui/stdio2 browser.
Though, we copy symbol_conf.show_nr_samples to
annotation__default_options.show_nr_samples but that is not really
effective as we don't use annotation__default_options once we copy
default options to dynamic variable annotate.opts in cmd_annotate().
Instead of all these complication, keep only one variable and use it all
over. symbol_conf.show_nr_samples is used by perf report/top as well. So
let's kill annotation_options.show_nr_samples.
On a side note, I've kept annotation_options.show_nr_samples definition
because it's still used by perf-config code. Follow up patch to fix
perf-config for annotate will remove annotation_options.show_nr_samples.
Ravi Bangoria [Thu, 13 Feb 2020 06:43:00 +0000 (12:13 +0530)]
perf annotate: Fix --show-total-period for tui/stdio2
perf annotate --show-total-period does not really show total period.
The reason is we have two separate variables for the same purpose.
One is in symbol_conf.show_total_period and another is
annotation_options.show_total_period.
We save command line option in symbol_conf.show_total_period but uses
annotation_option.show_total_period while rendering tui/stdio2 browser.
Though, we copy symbol_conf.show_total_period to
annotation__default_options.show_total_period but that is not really
effective as we don't use annotation__default_options once we copy
default options to dynamic variable annotate.opts in cmd_annotate().
Instead of all these complication, keep only one variable and use it all
over. symbol_conf.show_total_period is used by perf report/top as well.
So let's kill annotation_options.show_total_period.
On a side note, I've kept annotation_options.show_total_period
definition because it's still used by perf-config code. Follow up patch
to fix perf-config for annotate will remove
annotation_options.show_total_period.
Ravi Bangoria [Thu, 13 Feb 2020 06:42:59 +0000 (12:12 +0530)]
perf annotate/tui: Re-render title bar after switching back from script browser
The 'perf annotate' TUI browser provides a 'r' hot key to switch to a
script browser. But the annotate browser title bar becomes hidden while
switching back from script browser. Fix it.
tools headers UAPI: Update tools's copy of kvm.h headers
Picking the changes from:
5ef8acbdd687 ("KVM: nVMX: Emulate MTF when performing instruction emulation")
Silencing this perf build warning:
Warning: Kernel ABI header at 'tools/arch/x86/include/uapi/asm/kvm.h' differs from latest version at 'arch/x86/include/uapi/asm/kvm.h'
diff -u tools/arch/x86/include/uapi/asm/kvm.h arch/x86/include/uapi/asm/kvm.h
No change in tooling ensues, just the x86 kvm tooling gets rebuilt as
those headers are included in its build:
$ tools/perf/trace/beauty/tracepoints/x86_msr.sh > after
$ diff -u before after
$
To silence this perf build warning:
Warning: Kernel ABI header at 'tools/arch/x86/include/asm/msr-index.h' differs from latest version at 'arch/x86/include/asm/msr-index.h'
diff -u tools/arch/x86/include/asm/msr-index.h arch/x86/include/asm/msr-index.h
Merge tag 'devfreq-fixes-for-5.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/chanwoo/linux
Pull a devfreq fix for 5.6-rc4 from Chanwoo Choi:
"Revert "PM / devfreq: Modify the device name as devfreq(X) for sysfs"
- This changes as devfreq(X) cause break some user space applications
such as Android HAL from Unisoc and Hikey. In result, decide to revert it
for preventing the HAL layer problem."
* tag 'devfreq-fixes-for-5.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/chanwoo/linux:
Revert "PM / devfreq: Modify the device name as devfreq(X) for sysfs"
Vincent Guittot [Tue, 18 Feb 2020 14:45:34 +0000 (15:45 +0100)]
sched/fair: Fix statistics for find_idlest_group()
sgs->group_weight is not set while gathering statistics in
update_sg_wakeup_stats(). This means that a group can be classified as
fully busy with 0 running tasks if utilization is high enough.
cpufreq: Fix policy initialization for internal governor drivers
Before commit 1e4f63aecb53 ("cpufreq: Avoid creating excessively
large stack frames") the initial value of the policy field in struct
cpufreq_policy set by the driver's ->init() callback was implicitly
passed from cpufreq_init_policy() to cpufreq_set_policy() if the
default governor was neither "performance" nor "powersave". After
that commit, however, cpufreq_init_policy() must take that case into
consideration explicitly and handle it as appropriate, so make that
happen.
Aaro Koskinen [Wed, 26 Feb 2020 16:49:01 +0000 (18:49 +0200)]
net: stmmac: fix notifier registration
We cannot register the same netdev notifier multiple times when probing
stmmac devices. Register the notifier only once in module init, and also
make debugfs creation/deletion safe against simultaneous notifier call.
Fixes: 481a7d154cbb ("stmmac: debugfs entry name is not be changed when udev rename device name.") Signed-off-by: Aaro Koskinen <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Antoine Tenart [Wed, 26 Feb 2020 15:26:50 +0000 (16:26 +0100)]
net: phy: mscc: fix firmware paths
The firmware paths for the VSC8584 PHYs not not contain the leading
'microchip/' directory, as used in linux-firmware, resulting in an
error when probing the driver. This patch fixes it.
Fixes: a5afc1678044 ("net: phy: mscc: add support for VSC8584 PHY") Signed-off-by: Antoine Tenart <[email protected]> Reviewed-by: Andrew Lunn <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Address the issue adding a dummy icsk_sync_mss callback.
To properly sync the subflows mss and options list we need some
additional infrastructure, which will land to net-next.
Sudheesh Mavila [Wed, 26 Feb 2020 07:10:45 +0000 (12:40 +0530)]
net: phy: corrected the return value for genphy_check_and_restart_aneg and genphy_c45_check_and_restart_aneg
When auto-negotiation is not required, return value should be zero.
Changes v1->v2:
- improved comments and code as Andrew Lunn and Heiner Kallweit suggestion
- fixed issue in genphy_c45_check_and_restart_aneg as Russell King
suggestion.
Fixes: 2a10ab043ac5 ("net: phy: add genphy_check_and_restart_aneg()") Fixes: 1af9f16840e9 ("net: phy: add genphy_c45_check_and_restart_aneg()") Signed-off-by: Sudheesh Mavila <[email protected]> Reviewed-by: Heiner Kallweit <[email protected]> Reviewed-by: Andrew Lunn <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Ursula Braun [Tue, 25 Feb 2020 15:34:36 +0000 (16:34 +0100)]
net/smc: fix cleanup for linkgroup setup failures
If an SMC connection to a certain peer is setup the first time,
a new linkgroup is created. In case of setup failures, such a
linkgroup is unusable and should disappear. As a first step the
linkgroup is removed from the linkgroup list in smc_lgr_forget().
There are 2 problems:
smc_listen_decline() might be called before linkgroup creation
resulting in a crash due to calling smc_lgr_forget() with
parameter NULL.
If a setup failure occurs after linkgroup creation, the connection
is never unregistered from the linkgroup, preventing linkgroup
freeing.
This patch introduces an enhanced smc_lgr_cleanup_early() function
which
* contains a linkgroup check for early smc_listen_decline()
invocations
* invokes smc_conn_free() to guarantee unregistering of the
connection.
* schedules fast linkgroup removal of the unusable linkgroup
And the unused function smcd_conn_free() is removed from smc_core.h.
Fixes: 3b2dec2603d5b ("net/smc: restructure client and server code in af_smc") Fixes: 2a0674fffb6bc ("net/smc: improve abnormal termination of link groups") Signed-off-by: Ursula Braun <[email protected]> Signed-off-by: Karsten Graul <[email protected]> Signed-off-by: David S. Miller <[email protected]>
net: bcmgenet: Clear ID_MODE_DIS in EXT_RGMII_OOB_CTRL when not needed
Outdated Raspberry Pi 4 firmware might configure the external PHY as
rgmii although the kernel currently sets it as rgmii-rxid. This makes
connections unreliable as ID_MODE_DIS is left enabled. To avoid this,
explicitly clear that bit whenever we don't need it.
Fixes: da38802211cc ("net: bcmgenet: Add RGMII_RXID support") Signed-off-by: Nicolas Saenz Julienne <[email protected]> Acked-by: Florian Fainelli <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Jiri Pirko [Tue, 25 Feb 2020 12:54:12 +0000 (13:54 +0100)]
sched: act: count in the size of action flags bitfield
The put of the flags was added by the commit referenced in fixes tag,
however the size of the message was not extended accordingly.
Fix this by adding size of the flags bitfield to the message size.
Fixes: e38226786022 ("net: sched: update action implementations to support flags") Signed-off-by: Jiri Pirko <[email protected]> Signed-off-by: David S. Miller <[email protected]>
net: core: devlink.c: Use built-in RCU list checking
list_for_each_entry_rcu() has built-in RCU and lock checking.
Pass cond argument to list_for_each_entry_rcu() to silence
false lockdep warning when CONFIG_PROVE_RCU_LIST is enabled.
The devlink->lock is held when devlink_dpipe_table_find()
is called in non RCU read side section. Therefore, pass struct devlink
to devlink_dpipe_table_find() for lockdep checking.
Florian Fainelli [Mon, 24 Feb 2020 23:56:32 +0000 (15:56 -0800)]
net: dsa: bcm_sf2: Forcibly configure IMP port for 1Gb/sec
We are still experiencing some packet loss with the existing advanced
congestion buffering (ACB) settings with the IMP port configured for
2Gb/sec, so revert to conservative link speeds that do not produce
packet loss until this is resolved.
Fixes: 8f1880cbe8d0 ("net: dsa: bcm_sf2: Configure IMP port for 2Gb/sec") Fixes: de34d7084edd ("net: dsa: bcm_sf2: Only 7278 supports 2Gb/sec IMP port") Signed-off-by: Florian Fainelli <[email protected]> Reviewed-by: Vivien Didelot <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Linus Torvalds [Wed, 26 Feb 2020 23:54:52 +0000 (15:54 -0800)]
Merge tag 'tag-chrome-platform-fixes-for-v5.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/chrome-platform/linux
Pull chrome platform fix from Benson Leung:
"Fix a build warning"
* tag 'tag-chrome-platform-fixes-for-v5.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/chrome-platform/linux:
platform/chrome: wilco_ec: Include asm/unaligned instead of linux/ path
Cong Wang [Thu, 13 Feb 2020 06:53:52 +0000 (22:53 -0800)]
netfilter: xt_hashlimit: unregister proc file before releasing mutex
Before releasing the global mutex, we only unlink the hashtable
from the hash list, its proc file is still not unregistered at
this point. So syzbot could trigger a race condition where a
parallel htable_create() could register the same file immediately
after the mutex is released.
Move htable_remove_proc_entry() back to mutex protection to
fix this. And, fold htable_destroy() into htable_put() to make
the code slightly easier to understand.
Reported-and-tested-by: [email protected] Fixes: c4a3922d2d20 ("netfilter: xt_hashlimit: reduce hashlimit_mutex scope for htable_put()") Signed-off-by: Cong Wang <[email protected]> Signed-off-by: Pablo Neira Ayuso <[email protected]>
Michal Kubecek [Mon, 24 Feb 2020 19:42:12 +0000 (20:42 +0100)]
ethtool: limit bitset size
Syzbot reported that ethnl_compact_sanity_checks() can be tricked into
reading past the end of ETHTOOL_A_BITSET_VALUE and ETHTOOL_A_BITSET_MASK
attributes and even the message by passing a value between (u32)(-31)
and (u32)(-1) as ETHTOOL_A_BITSET_SIZE.
The problem is that DIV_ROUND_UP(attr_nbits, 32) is 0 for such values so
that zero length ETHTOOL_A_BITSET_VALUE will pass the length check but
ethnl_bitmap32_not_zero() check would try to access up to 512 MB of
attribute "payload".
Prevent this overflow byt limiting the bitset size. Technically, compact
bitset format would allow bitset sizes up to almost 2^18 (so that the
nest size does not exceed U16_MAX) but bitsets used by ethtool are much
shorter. S16_MAX, the largest value which can be directly used as an
upper limit in policy, should be a reasonable compromise.
Amritha Nambiar [Mon, 24 Feb 2020 18:56:00 +0000 (10:56 -0800)]
net: Fix Tx hash bound checking
Fixes the lower and upper bounds when there are multiple TCs and
traffic is on the the same TC on the same device.
The lower bound is represented by 'qoffset' and the upper limit for
hash value is 'qcount + qoffset'. This gives a clean Rx to Tx queue
mapping when there are multiple TCs, as the queue indices for upper TCs
will be offset by 'qoffset'.
v2: Fixed commit description based on comments.
Fixes: 1b837d489e06 ("net: Revoke export for __skb_tx_hash, update it to just be static skb_tx_hash") Fixes: eadec877ce9c ("net: Add support for subordinate traffic classes to netdev_pick_tx") Signed-off-by: Amritha Nambiar <[email protected]> Reviewed-by: Alexander Duyck <[email protected]> Reviewed-by: Sridhar Samudrala <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Daniel Vetter [Sat, 22 Feb 2020 17:54:32 +0000 (18:54 +0100)]
drm/radeon: Inline drm_get_pci_dev
It's the last user, and more importantly, it's the last non-legacy
user of anything in drm_pci.c.
The only tricky bit is the agp initialization. But a close look shows
that radeon does not use the drm_agp midlayer (the main use of that is
drm_bufs for legacy drivers), and instead could use the agp subsystem
directly (like nouveau does already). Hence we can just pull this in
too.
A further step would be to entirely drop the use of drm_device->agp,
but feels like too much churn just for this patch.
drm/amdgpu: drop deprecated drm_get_pci_dev and drm_put_dev
No idea whether this was intentional or accidental breakage, but I
guess anyone who manages to boot a this modern gpu behind an agp
bridge deserves a price. A price I never expect anyone to ever collect
:-)
Linus Torvalds [Wed, 26 Feb 2020 18:34:42 +0000 (10:34 -0800)]
Merge tag 'trace-v5.6-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace
Pull tracing and bootconfig updates:
"Fixes and changes to bootconfig before it goes live in a release.
Change in API of bootconfig (before it comes live in a release):
- Have a magic value "BOOTCONFIG" in initrd to know a bootconfig
exists
- Set CONFIG_BOOT_CONFIG to 'n' by default
- Show error if "bootconfig" on cmdline but not compiled in
- Prevent redefining the same value
- Have a way to append values
- Added a SELECT BLK_DEV_INITRD to fix a build failure
Synthetic event fixes:
- Switch to raw_smp_processor_id() for recording CPU value in preempt
section. (No care for what the value actually is)
- Fix samples always recording u64 values
- Fix endianess
- Check number of values matches number of fields
- Fix a printing bug
Fix of trace_printk() breaking postponed start up tests
Make a function static that is only used in a single file"
* tag 'trace-v5.6-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace:
bootconfig: Fix CONFIG_BOOTTIME_TRACING dependency issue
bootconfig: Add append value operator support
bootconfig: Prohibit re-defining value on same key
bootconfig: Print array as multiple commands for legacy command line
bootconfig: Reject subkey and value on same parent key
tools/bootconfig: Remove unneeded error message silencer
bootconfig: Add bootconfig magic word for indicating bootconfig explicitly
bootconfig: Set CONFIG_BOOT_CONFIG=n by default
tracing: Clear trace_state when starting trace
bootconfig: Mark boot_config_checksum() static
tracing: Disable trace_printk() on post poned tests
tracing: Have synthetic event test use raw_smp_processor_id()
tracing: Fix number printing bug in print_synth_event()
tracing: Check that number of vals matches number of synth event fields
tracing: Make synth_event trace functions endian-correct
tracing: Make sure synth_event_trace() example always uses u64
Linus Torvalds [Wed, 26 Feb 2020 18:28:59 +0000 (10:28 -0800)]
Merge tag 'linux-kselftest-kunit-5.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest
Pull Kunit fixes from Shuah Khan:
"This Kselftest kunit update consists of fixes to documentation and
the run-time tool from Brendan Higgins and Heidi Fahim"
* tag 'linux-kselftest-kunit-5.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest:
kunit: run kunit_tool from any directory
kunit: test: Improve error messages for kunit_tool when kunitconfig is invalid
Documentation: kunit: fixed sphinx error in code block
Using the -Werror flag breaks the build for me due to mostly harmless
KASAN or similar warnings:
arch/x86/kvm/x86.c: In function ‘kvm_timer_init’:
arch/x86/kvm/x86.c:7209:1: error: the frame size of 1112 bytes is larger than 1024 bytes [-Werror=frame-larger-than=]
Feel free to add a CONFIG_WERROR if you care strong enough, but don't
break peoples builds for absolutely no good reason.
Linus Torvalds [Mon, 24 Feb 2020 20:47:14 +0000 (12:47 -0800)]
signal: avoid double atomic counter increments for user accounting
When queueing a signal, we increment both the users count of pending
signals (for RLIMIT_SIGPENDING tracking) and we increment the refcount
of the user struct itself (because we keep a reference to the user in
the signal structure in order to correctly account for it when freeing).
That turns out to be fairly expensive, because both of them are atomic
updates, and particularly under extreme signal handling pressure on big
machines, you can get a lot of cache contention on the user struct.
That can then cause horrid cacheline ping-pong when you do these
multiple accesses.
So change the reference counting to only pin the user for the _first_
pending signal, and to unpin it when the last pending signal is
dequeued. That means that when a user sees a lot of concurrent signal
queuing - which is the only situation when this matters - the only
atomic access needed is generally the 'sigpending' count update.
This was noticed because of a particularly odd timing artifact on a
dual-socket 96C/192T Cascade Lake platform: when you get into bad
contention, on that machine for some reason seems to be much worse when
the contention happens in the upper 32-byte half of the cacheline.
As a result, the kernel test robot will-it-scale 'signal1' benchmark had
an odd performance regression simply due to random alignment of the
'struct user_struct' (and pointed to a completely unrelated and
apparently nonsensical commit for the regression).
Avoiding the double increments (and decrements on the dequeueing side,
of course) makes for much less contention and hugely improved
performance on that will-it-scale microbenchmark.
Quoting Feng Tang:
"It makes a big difference, that the performance score is tripled! bump
from original 17000 to 54000. Also the gap between 5.0-rc6 and
5.0-rc6+Jiri's patch is reduced to around 2%"
[ The "2% gap" is the odd cacheline placement difference on that
platform: under the extreme contention case, the effect of which half
of the cacheline was hot was 5%, so with the reduced contention the
odd timing artifact is reduced too ]
It does help in the non-contended case too, but is not nearly as
noticeable.
Turns out that this is due to percpu_ref_switch_to_atomic() only
grabbing a reference to the percpu refcount if it's not already in
atomic mode. io_uring drops a ref and re-gets it when switching back to
percpu mode. We attempt to protect against this with the FFD_F_ATOMIC
bit, but that isn't reliable.
We don't actually need to juggle these refcounts between atomic and
percpu switch, we can just do them when we've switched to atomic mode.
This removes the need for FFD_F_ATOMIC, which wasn't reliable.
Fixes: 05f3fb3c5397 ("io_uring: avoid ring quiesce for fixed file set unregister and update") Reported-by: Dan Melnic <[email protected]> Signed-off-by: Jens Axboe <[email protected]>
Reviewed-by: Johannes Thumshirn <[email protected]> Reviewed-by: Hannes Reinecke <[email protected]> Signed-off-by: John Garry <[email protected]>
--
This minor patch had been carried as part of the blk-mq shared tags RFC,
I'd rather not carry it anymore as it required rebasing, so now or never.. Signed-off-by: Jens Axboe <[email protected]>
Masahiro Yamada [Sat, 22 Feb 2020 19:04:33 +0000 (04:04 +0900)]
kbuild: add dtbs_check to PHONY
The dtbs_check should be a phony target, but currently it is not
specified so.
'make dtbs_check' works even if a file named 'dtbs_check' exists
because it depends on another phony target, scripts_dtc, but we
should not rely on it.
Masahiro Yamada [Sat, 22 Feb 2020 19:04:31 +0000 (04:04 +0900)]
kbuild: fix DT binding schema rule to detect command line changes
This if_change_rule is not working properly; it cannot detect any
command line change.
The reason is because cmd-check in scripts/Kbuild.include compares
$(cmd_$@) and $(cmd_$1), but cmd_dtc_dt_yaml does not exist here.
For if_change_rule to work properly, the stem part of cmd_* and rule_*
must match. Because this cmd_and_fixdep invokes cmd_dtc, this rule must
be named rule_dtc.
Fixes: 4f0e3a57d6eb ("kbuild: Add support for DT binding schema checks") Signed-off-by: Masahiro Yamada <[email protected]> Acked-by: Rob Herring <[email protected]>
Masahiro Yamada [Wed, 19 Feb 2020 01:15:19 +0000 (10:15 +0900)]
kbuild: remove wrong documentation about mandatory-y
This sentence does not make sense in the section about mandatory-y.
This seems to be a copy-paste mistake of commit fcc8487d477a ("uapi:
export all headers under uapi directories").
The correct description would be "The convention is to list one
mandatory-y per line ...".
I just removed it instead of fixing it. If such information is needed,
it could be commented in include/asm-generic/Kbuild and
include/uapi/asm-generic/Kbuild.
This function is consistent with using size instead of seed->size
(except for one place that this patch fixes), but it reads seed->size
without using READ_ONCE, which means the compiler might still do
something unwanted. So, this commit simply adds the READ_ONCE
wrapper.
Ard Biesheuvel [Fri, 21 Feb 2020 08:48:48 +0000 (09:48 +0100)]
efi/x86: Handle by-ref arguments covering multiple pages in mixed mode
The mixed mode runtime wrappers are fragile when it comes to how the
memory referred to by its pointer arguments are laid out in memory, due
to the fact that it translates these addresses to physical addresses that
the runtime services can dereference when running in 1:1 mode. Since
vmalloc'ed pages (including the vmap'ed stack) are not contiguous in the
physical address space, this scheme only works if the referenced memory
objects do not cross page boundaries.
Currently, the mixed mode runtime service wrappers require that all by-ref
arguments that live in the vmalloc space have a size that is a power of 2,
and are aligned to that same value. While this is a sensible way to
construct an object that is guaranteed not to cross a page boundary, it is
overly strict when it comes to checking whether a given object violates
this requirement, as we can simply take the physical address of the first
and the last byte, and verify that they point into the same physical page.
When this check fails, we emit a WARN(), but then simply proceed with the
call, which could cause data corruption if the next physical page belongs
to a mapping that is entirely unrelated.
Given that with vmap'ed stacks, this condition is much more likely to
trigger, let's relax the condition a bit, but fail the runtime service
call if it does trigger.
Ard Biesheuvel [Fri, 21 Feb 2020 08:48:47 +0000 (09:48 +0100)]
efi/x86: Remove support for EFI time and counter services in mixed mode
Mixed mode calls at runtime are rather tricky with vmap'ed stacks,
as we can no longer assume that data passed in by the callers of the
EFI runtime wrapper routines is contiguous in physical memory.
We need to fix this, but before we do, let's drop the implementations
of routines that we know are never used on x86, i.e., the RTC related
ones. Given that UEFI rev2.8 permits any runtime service to return
EFI_UNSUPPORTED at runtime, let's return that instead.
As get_next_high_mono_count() is never used at all, even on other
architectures, let's make that return EFI_UNSUPPORTED too.
Ard Biesheuvel [Fri, 21 Feb 2020 08:48:46 +0000 (09:48 +0100)]
efi/x86: Align GUIDs to their size in the mixed mode runtime wrapper
Hans reports that his mixed mode systems running v5.6-rc1 kernels hit
the WARN_ON() in virt_to_phys_or_null_size(), caused by the fact that
efi_guid_t objects on the vmap'ed stack happen to be misaligned with
respect to their sizes. As a quick (i.e., backportable) fix, copy GUID
pointer arguments to the local stack into a buffer that is naturally
aligned to its size, so that it is guaranteed to cover only one
physical page.
Note that on x86, we cannot rely on the stack pointer being aligned
the way the compiler expects, so we need to allocate an 8-byte aligned
buffer of sufficient size, and copy the GUID into that buffer at an
offset that is aligned to 16 bytes.
Jens Axboe [Wed, 26 Feb 2020 00:48:55 +0000 (17:48 -0700)]
io_uring: import_single_range() returns 0/-ERROR
Unlike the other core import helpers, import_single_range() returns 0 on
success, not the length imported. This means that links that depend on
the result of non-vec based IORING_OP_{READ,WRITE} that were added for
5.5 get errored when they should not be.
Jens Axboe [Tue, 25 Feb 2020 20:25:41 +0000 (13:25 -0700)]
io_uring: pick up link work on submit reference drop
If work completes inline, then we should pick up a dependent link item
in __io_queue_sqe() as well. If we don't do so, we're forced to go async
with that item, which is suboptimal.
This also fixes an issue with io_put_req_find_next(), which always looks
up the next work item. That should only be done if we're dropping the
last reference to the request, to prevent multiple lookups of the same
work item.
Outside of being a fix, this also enables a good cleanup series for 5.7,
where we never have to pass 'nxt' around or into the work handlers.
Stefano Brivio [Fri, 21 Feb 2020 02:04:22 +0000 (03:04 +0100)]
selftests: nft_concat_range: Add test for reported add/flush/add issue
Add a specific test for the crash reported by Phil Sutter and addressed
in the previous patch. The test cases that, in my intention, should
have covered these cases, that is, the ones from the 'concurrency'
section, don't run these sequences tightly enough and spectacularly
failed to catch this.
While at it, define a convenient way to add these kind of tests, by
adding a "reported issues" test section.
It's more convenient, for this particular test, to execute the set
setup in its own function. However, future test cases like this one
might need to call setup functions, and will typically need no tools
other than nft, so allow for this in check_tools().
The original form of the reproducer used here was provided by Phil.
Stefano Brivio [Fri, 21 Feb 2020 02:04:21 +0000 (03:04 +0100)]
nft_set_pipapo: Actually fetch key data in nft_pipapo_remove()
Phil reports that adding elements, flushing and re-adding them
right away:
nft add table t '{ set s { type ipv4_addr . inet_service; flags interval; }; }'
nft add element t s '{ 10.0.0.1 . 22-25, 10.0.0.1 . 10-20 }'
nft flush set t s
nft add element t s '{ 10.0.0.1 . 10-20, 10.0.0.1 . 22-25 }'
which is caused by dangling elements that have been deactivated, but
never removed.
On a flush operation, nft_pipapo_walk() walks through all the elements
in the mapping table, which are then deactivated by nft_flush_set(),
one by one, and added to the commit list for removal. Element data is
then freed.
On transaction commit, nft_pipapo_remove() is called, and failed to
remove these elements, leading to the stale references in the mapping.
The first symptom of this, revealed by KASan, is a one-byte
use-after-free in subsequent calls to nft_pipapo_walk(), which is
usually not enough to trigger a panic. When stale elements are used
more heavily, though, such as double-free via nft_pipapo_destroy()
as in Phil's case, the problem becomes more noticeable.
The issue comes from that fact that, on a flush operation,
nft_pipapo_remove() won't get the actual key data via elem->key,
elements to be deleted upon commit won't be found by the lookup via
pipapo_get(), and removal will be skipped. Key data should be fetched
via nft_set_ext_key(), instead.
Reported-by: Phil Sutter <[email protected]> Fixes: 3c4287f62044 ("nf_tables: Add set type for arbitrary concatenation of ranges") Signed-off-by: Stefano Brivio <[email protected]> Signed-off-by: Pablo Neira Ayuso <[email protected]>
Merge branch 'master' of git://blackhole.kfki.hu/nf
Jozsef Kadlecsik says:
====================
ipset patches for nf
The first one is larger than usual, but the issue could not be solved simpler.
Also, it's a resend of the patch I submitted a few days ago, with a one line
fix on top of that: the size of the comment extensions was not taken into
account at reporting the full size of the set.
- Fix "INFO: rcu detected stall in hash_xxx" reports of syzbot
by introducing region locking and using workqueue instead of timer based
gc of timed out entries in hash types of sets in ipset.
- Fix the forceadd evaluation path - the bug was also uncovered by the syzbot.
====================
Chris Wilson [Fri, 21 Feb 2020 22:18:18 +0000 (22:18 +0000)]
drm/i915: Avoid recursing onto active vma from the shrinker
We mark the vma as active while binding it in order to protect outselves
from being shrunk under mempressure. This only works if we are strict in
not attempting to shrink active objects.
Michał Winiarski [Wed, 19 Feb 2020 16:18:22 +0000 (17:18 +0100)]
drm/i915/pmu: Avoid using globals for PMU events
Attempting to bind / unbind module from devices where we have both
integrated and discreete GPU handled by i915, will cause us to try and
double free the global state, hitting null ptr deref in free_event_attributes.
Michał Winiarski [Wed, 19 Feb 2020 16:18:21 +0000 (17:18 +0100)]
drm/i915/pmu: Avoid using globals for CPU hotplug state
Attempting to bind / unbind module from devices where we have both
integrated and discreete GPU handled by i915 can lead to leaks and
warnings from cpuhp:
Error: Removing state XXX which has instances left.
Jani Nikula [Fri, 21 Feb 2020 10:54:14 +0000 (12:54 +0200)]
drm/i915: fix header test with GCOV
$(CC) with $(CFLAGS_GCOV) assumes the output filename with .gcno suffix
appended is writable. This is not the case when the output filename is
/dev/null:
IOCB_NOWAIT can't just be ignored as it breaks applications expecting
it not to block. Just refuse the operation as applications must handle
that (e.g. by falling back to a thread pool).
Fixes: 8dcc1a9d90c1 ("fs: New zonefs file system") Signed-off-by: Christoph Hellwig <[email protected]> Signed-off-by: Damien Le Moal <[email protected]>
Since commit d8a953ddde5e ("bootconfig: Set CONFIG_BOOT_CONFIG=n by
default") also changed the CONFIG_BOOTTIME_TRACING to select
CONFIG_BOOT_CONFIG to show the boot-time tracing on the menu,
it introduced wrong dependencies with BLK_DEV_INITRD as below.
WARNING: unmet direct dependencies detected for BOOT_CONFIG
Depends on [n]: BLK_DEV_INITRD [=n]
Selected by [y]:
- BOOTTIME_TRACING [=y] && TRACING_SUPPORT [=y] && FTRACE [=y] && TRACING [=y]
This makes the CONFIG_BOOT_CONFIG selects CONFIG_BLK_DEV_INITRD to
fix this error and make CONFIG_BOOTTIME_TRACING=n by default, so
that both boot-time tracing and boot configuration off but those
appear on the menu list.
Jens Axboe [Tue, 25 Feb 2020 18:52:56 +0000 (11:52 -0700)]
io-wq: ensure work->task_pid is cleared on init
We use ->task_pid for exit cancellation, but we need to ensure it's
cleared to zero for io_req_work_grab_env() to do the right thing. Take
a suggestion from Bart and clear the whole thing, just setting the
function passed in. This makes it more future proof as well.
Fixes: 36282881a795 ("io-wq: add io_wq_cancel_pid() to cancel based on a specific pid") Signed-off-by: Jens Axboe <[email protected]>
icmp: allow icmpv6_ndo_send to work with CONFIG_IPV6=n
The icmpv6_send function has long had a static inline implementation
with an empty body for CONFIG_IPV6=n, so that code calling it doesn't
need to be ifdef'd. The new icmpv6_ndo_send function, which is intended
for drivers as a drop-in replacement with an identical function
signature, should follow the same pattern. Without this patch, drivers
that used to work with CONFIG_IPV6=n now result in a linker error.
Cc: Chen Zhou <[email protected]> Reported-by: Hulk Robot <[email protected]> Fixes: 0b41713b6066 ("icmp: introduce helper for nat'd source address in network device context") Signed-off-by: Jason A. Donenfeld <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Linus Torvalds [Tue, 25 Feb 2020 18:14:39 +0000 (10:14 -0800)]
Merge tag 'riscv-for-linux-5.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/riscv/linux
Pull RISC-V fixes from Palmer Dabbelt:
"This contains a handful of RISC-V related fixes that I've collected
and would like to target for 5.6-rc4:
- A fix to set up the PMPs on boot, which allows the kernel to access
memory on systems that don't set up permissive PMPs before getting
to Linux. This only effects machine-mode kernels, which currently
means only NOMMU kernels.
- A fix to avoid enabling supervisor-mode interrupts when running in
machine-mode, also only for NOMMU kernels.
- A pair of fixes to our KASAN support to avoid corrupting memory.
- A gitignore fix.
This boots on QEMU's virt board for me"
* tag 'riscv-for-linux-5.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/riscv/linux:
riscv: adjust the indent
riscv: allocate a complete page size for each page table
riscv: Fix gitignore
RISC-V: Don't enable all interrupts in trap_init()
riscv: set pmp configuration if kernel is running in M-mode
Linus Torvalds [Tue, 25 Feb 2020 18:09:41 +0000 (10:09 -0800)]
Merge branch 'mips-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux
Pull MIPS fixes from Paul Burton:
"Here are a few MIPS fixes, and a MAINTAINERS update to hand over MIPS
maintenance to Thomas Bogendoerfer - this will be my final pull
request as MIPS maintainer.
Thanks for your helpful comments, useful corrections & responsiveness
during the time I've fulfilled the role, and I'm sure I'll pop up
elsewhere in the tree somewhere down the line"
* 'mips-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux:
MAINTAINERS: Hand MIPS over to Thomas
MIPS: ingenic: DTS: Fix watchdog nodes
MIPS: X1000: Fix clock of watchdog node.
MIPS: vdso: Wrap -mexplicit-relocs in cc-option
MIPS: VPE: Fix a double free and a memory leak in 'release_vpe()'
MIPS: cavium_octeon: Fix syncw generation.
mips: vdso: add build time check that no 'jalr t9' calls left
MIPS: Disable VDSO time functionality on microMIPS
mips: vdso: fix 'jalr t9' crash in vdso code
Jens Axboe [Tue, 25 Feb 2020 15:47:30 +0000 (08:47 -0700)]
io-wq: remove spin-for-work optimization
Andres reports that buffered IO seems to suck up more cycles than we
would like, and he narrowed it down to the fact that the io-wq workers
will briefly spin for more work on completion of a work item. This was
a win on the networking side, but apparently some other cases take a
hit because of it. Remove the optimization to avoid burning more CPU
than we have to for disk IO.
Xiaoguang Wang [Tue, 25 Feb 2020 14:12:08 +0000 (22:12 +0800)]
io_uring: fix poll_list race for SETUP_IOPOLL|SETUP_SQPOLL
After making ext4 support iopoll method:
let ext4_file_operations's iopoll method be iomap_dio_iopoll(),
we found fio can easily hang in fio_ioring_getevents() with below fio
job:
rm -f testfile; sync;
sudo fio -name=fiotest -filename=testfile -iodepth=128 -thread
-rw=write -ioengine=io_uring -hipri=1 -sqthread_poll=1 -direct=1
-bs=4k -size=10G -numjobs=8 -runtime=2000 -group_reporting
with IORING_SETUP_SQPOLL and IORING_SETUP_IOPOLL enabled.
There are two issues that results in this hang, one reason is that
when IORING_SETUP_SQPOLL and IORING_SETUP_IOPOLL are enabled, fio
does not use io_uring_enter to get completed events, it relies on
kernel io_sq_thread to poll for completed events.
Another reason is that there is a race: when io_submit_sqes() in
io_sq_thread() submits a batch of sqes, variable 'inflight' will
record the number of submitted reqs, then io_sq_thread will poll for
reqs which have been added to poll_list. But note, if some previous
reqs have been punted to io worker, these reqs will won't be in
poll_list timely. io_sq_thread() will only poll for a part of previous
submitted reqs, and then find poll_list is empty, reset variable
'inflight' to be zero. If app just waits these deferred reqs and does
not wake up io_sq_thread again, then hang happens.
For app that entirely relies on io_sq_thread to poll completed requests,
let io_iopoll_req_issued() wake up io_sq_thread properly when adding new
element to poll_list, and when io_sq_thread prepares to sleep, check
whether poll_list is empty again, if not empty, continue to poll.
Jan Kara [Thu, 6 Feb 2020 14:28:12 +0000 (15:28 +0100)]
blktrace: Protect q->blk_trace with RCU
KASAN is reporting that __blk_add_trace() has a use-after-free issue
when accessing q->blk_trace. Indeed the switching of block tracing (and
thus eventual freeing of q->blk_trace) is completely unsynchronized with
the currently running tracing and thus it can happen that the blk_trace
structure is being freed just while __blk_add_trace() works on it.
Protect accesses to q->blk_trace by RCU during tracing and make sure we
wait for the end of RCU grace period when shutting down tracing. Luckily
that is rare enough event that we can afford that. Note that postponing
the freeing of blk_trace to an RCU callback should better be avoided as
it could have unexpected user visible side-effects as debugfs files
would be still existing for a short while block tracing has been shut
down.
Stefano Brivio [Fri, 21 Feb 2020 02:11:56 +0000 (03:11 +0100)]
selftests: nft_concat_range: Move option for 'list ruleset' before command
Before nftables commit fb9cea50e8b3 ("main: enforce options before
commands"), 'nft list ruleset -a' happened to work, but it's wrong
and won't work anymore. Replace it by 'nft -a list ruleset'.
Reported-by: Chen Yi <[email protected]> Fixes: 611973c1e06f ("selftests: netfilter: Introduce tests for sets with range concatenation") Signed-off-by: Stefano Brivio <[email protected]> Signed-off-by: Pablo Neira Ayuso <[email protected]>
Kees Cook [Sat, 22 Feb 2020 00:02:39 +0000 (16:02 -0800)]
docs: Fix empty parallelism argument
When there was no parallelism (no top-level -j arg and a pre-1.7
sphinx-build), the argument passed would be empty ("") instead of just
being missing, which would (understandably) badly confuse sphinx-build.
Fix this by removing the quotes.
Paul Burton [Sat, 22 Feb 2020 17:04:17 +0000 (09:04 -0800)]
MAINTAINERS: Hand MIPS over to Thomas
My time with MIPS the company has reached its end, and so at best I'll
have little time spend on maintaining arch/mips/.
Ralf last authored a patch over 2 years ago, the last time he committed
one is even further back & activity was sporadic for a while before
that. The reality is that he isn't active.
Having a new maintainer with time to do things properly will be
beneficial all round. Thomas Bogendoerfer has been involved in MIPS
development for a long time & has offered to step up as maintainer, so
add Thomas and remove myself & Ralf from the MIPS entry.
Ralf already has an entry in CREDITS to honor his contributions, so this
just adds one for me.
Ming Lei [Tue, 25 Feb 2020 01:04:32 +0000 (09:04 +0800)]
blk-mq: insert passthrough request into hctx->dispatch directly
For some reason, device may be in one situation which can't handle
FS request, so STS_RESOURCE is always returned and the FS request
will be added to hctx->dispatch. However passthrough request may
be required at that time for fixing the problem. If passthrough
request is added to scheduler queue, there isn't any chance for
blk-mq to dispatch it given we prioritize requests in hctx->dispatch.
Then the FS IO request may never be completed, and IO hang is caused.
So passthrough request has to be added to hctx->dispatch directly
for fixing the IO hang.
Fix this issue by inserting passthrough request into hctx->dispatch
directly together withing adding FS request to the tail of
hctx->dispatch in blk_mq_dispatch_rq_list(). Actually we add FS request
to tail of hctx->dispatch at default, see blk_mq_request_bypass_insert().
Then it becomes consistent with original legacy IO request
path, in which passthrough request is always added to q->queue_head.
David S. Miller [Mon, 24 Feb 2020 23:43:38 +0000 (15:43 -0800)]
Merge tag 'mac80211-for-net-2020-02-24' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac80211
Johannes Berg
====================
A few fixes:
* remove a double mutex-unlock
* fix a leak in an error path
* NULL pointer check
* include if_vlan.h where needed
* avoid RCU list traversal when not under RCU
====================
Zong Li [Fri, 7 Feb 2020 09:52:44 +0000 (17:52 +0800)]
riscv: allocate a complete page size for each page table
Each page table should be created by allocating a complete page size
for it. Otherwise, the content of the page table would be corrupted
somewhere through memory allocation which allocates the memory at the
middle of the page table for other use.
Aurelien Aptel [Fri, 21 Feb 2020 10:19:06 +0000 (11:19 +0100)]
cifs: fix rename() by ensuring source handle opened with DELETE bit
To rename a file in SMB2 we open it with the DELETE access and do a
special SetInfo on it. If the handle is missing the DELETE bit the
server will fail the SetInfo with STATUS_ACCESS_DENIED.
We currently try to reuse any existing opened handle we have with
cifs_get_writable_path(). That function looks for handles with WRITE
access but doesn't check for DELETE, making rename() fail if it finds
a handle to reuse. Simple reproducer below.
To select handles with the DELETE bit, this patch adds a flag argument
to cifs_get_writable_path() and find_writable_file() and the existing
'bool fsuid_only' argument is converted to a flag.
The cifsFileInfo struct only stores the UNIX open mode but not the
original SMB access flags. Since the DELETE bit is not mapped in that
mode, this patch stores the access mask in cifs_fid on file open,
which is accessible from cifsFileInfo.
int main(int argc, char *argv[])
{
int fd, ret;
if (argc != 3) {
fprintf(stderr, "Usage: %s A B\n"
"create&open A in write mode, "
"rename A to B, close A\n", argv[0]);
return 0;
}
Ronnie Sahlberg [Tue, 18 Feb 2020 20:01:03 +0000 (06:01 +1000)]
cifs: don't leak -EAGAIN for stat() during reconnect
If from cifs_revalidate_dentry_attr() the SMB2/QUERY_INFO call fails with an
error, such as STATUS_SESSION_EXPIRED, causing the session to be reconnected
it is possible we will leak -EAGAIN back to the application even for
system calls such as stat() where this is not a valid error.
Fix this by re-trying the operation from within cifs_revalidate_dentry_attr()
if cifs_get_inode_info*() returns -EAGAIN.
This fixes stat() and possibly also other system calls that uses
cifs_revalidate_dentry*().
Adam Williamson [Wed, 19 Feb 2020 16:50:07 +0000 (17:50 +0100)]
scsi: compat_ioctl: cdrom: Replace .ioctl with .compat_ioctl in four appropriate places
Arnd Bergmann inadvertently typoed these in d320a9551e394 and 64cbfa96551a;
they seem to be the cause of
https://bugzilla.redhat.com/show_bug.cgi?id=1801353 , invalid SCSI commands
when udev tries to query a DVD drive.
[arnd] Found another instance of the same bug, also introduced in my
compat_ioctl series.
Linus Torvalds [Mon, 24 Feb 2020 19:40:23 +0000 (11:40 -0800)]
Merge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Pull crypto fixes from Herbert Xu:
"This fixes a Kconfig-related build error and an integer overflow in
chacha20poly1305"
* 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
crypto: chacha20poly1305 - prevent integer overflow on large input
tee: amdtee: amdtee depends on CRYPTO_DEV_CCP_DD
Linus Torvalds [Fri, 21 Feb 2020 20:43:35 +0000 (12:43 -0800)]
floppy: check FDC index for errors before assigning it
Jordy Zomer reported a KASAN out-of-bounds read in the floppy driver in
wait_til_ready().
Which on the face of it can't happen, since as Willy Tarreau points out,
the function does no particular memory access. Except through the FDCS
macro, which just indexes a static allocation through teh current fdc,
which is always checked against N_FDC.
Except the checking happens after we've already assigned the value.
The floppy driver is a disgrace (a lot of it going back to my original
horrd "design"), and has no real maintainer. Nobody has the hardware,
and nobody really cares. But it still gets used in virtual environment
because it's one of those things that everybody supports.
The whole thing should be re-written, or at least parts of it should be
seriously cleaned up. The 'current fdc' index, which is used by the
FDCS macro, and which is often shadowed by a local 'fdc' variable, is a
prime example of how not to write code.
But because nobody has the hardware or the motivation, let's just fix up
the immediate problem with a nasty band-aid: test the fdc index before
actually assigning it to the static 'fdc' variable.
net: bridge: fix stale eth hdr pointer in br_dev_xmit
In br_dev_xmit() we perform vlan filtering in br_allowed_ingress() but
if the packet has the vlan header inside (e.g. bridge with disabled
tx-vlan-offload) then the vlan filtering code will use skb_vlan_untag()
to extract the vid before filtering which in turn calls pskb_may_pull()
and we may end up with a stale eth pointer. Moreover the cached eth header
pointer will generally be wrong after that operation. Remove the eth header
caching and just use eth_hdr() directly, the compiler does the right thing
and calculates it only once so we don't lose anything.
Fixes: 057658cb33fb ("bridge: suppress arp pkts on BR_NEIGH_SUPPRESS ports") Signed-off-by: Nikolay Aleksandrov <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Esben Haabendal [Fri, 21 Feb 2020 06:47:58 +0000 (07:47 +0100)]
net: ll_temac: Handle DMA halt condition caused by buffer underrun
The SDMA engine used by TEMAC halts operation when it has finished
processing of the last buffer descriptor in the buffer ring.
Unfortunately, no interrupt event is generated when this happens,
so we need to setup another mechanism to make sure DMA operation is
restarted when enough buffers have been added to the ring.
Fixes: 92744989533c ("net: add Xilinx ll_temac device driver") Signed-off-by: Esben Haabendal <[email protected]> Signed-off-by: David S. Miller <[email protected]>
In order to handle such failures more graceful, this change splits the
receive loop into one for consuming the received buffers, and one for
allocating new buffers.
When GFP_ATOMIC allocations fail, the receive will continue with the
buffers that is still there, and with the expectation that the allocations
will succeed in a later call to receive.
Fixes: 92744989533c ("net: add Xilinx ll_temac device driver") Signed-off-by: Esben Haabendal <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Esben Haabendal [Fri, 21 Feb 2020 06:47:21 +0000 (07:47 +0100)]
net: ll_temac: Fix race condition causing TX hang
It is possible that the interrupt handler fires and frees up space in
the TX ring in between checking for sufficient TX ring space and
stopping the TX queue in temac_start_xmit. If this happens, the
queue wake from the interrupt handler will occur before the queue is
stopped, causing a lost wakeup and the adapter's transmit hanging.
To avoid this, after stopping the queue, check again whether there is
sufficient space in the TX ring. If so, wake up the queue again.
This is a port of the similar fix in axienet driver,
commit 7de44285c1f6 ("net: axienet: Fix race condition causing TX hang").
Fixes: 23ecc4bde21f ("net: ll_temac: fix checksum offload logic") Signed-off-by: Esben Haabendal <[email protected]> Signed-off-by: David S. Miller <[email protected]>
projects / linux.git / log