commit edc4746f25 upstream.
A recent fix addressing a deadlock on disconnect introduced a new bug
by moving the present flag out of the critical section protected by the
driver-data mutex. This could lead to a racing release() freeing the
driver data before disconnect() is done with it.
Due to insufficient locking a related use-after-free could be triggered
also before the above mentioned commit. Specifically, the driver needs
to hold the driver-data mutex also while checking the opened flag at
disconnect().
Fixes: c468a8aa79 ("usb: iowarrior: fix deadlock on disconnect")
Fixes: 946b960d13 ("USB: add driver for iowarrior devices.")
Cc: stable <stable@vger.kernel.org> # 2.6.21
Reported-by: syzbot+0761012cebf7bdb38137@syzkaller.appspotmail.com
Signed-off-by: Johan Hovold <johan@kernel.org>
Link: https://lore.kernel.org/r/20191009104846.5925-2-johan@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit b2fa7baee7 upstream.
The driver was using its struct usb_device pointer as an inverted
disconnected flag, but was setting it to NULL before making sure all
completion handlers had run. This could lead to a NULL-pointer
dereference in a number of dev_dbg statements in the completion handlers
which relies on said pointer.
The pointer was also dereferenced unconditionally in a dev_dbg statement
release() something which would lead to a NULL-deref whenever a device
was disconnected before the final character-device close if debugging
was enabled.
Fix this by unconditionally stopping all I/O and preventing
resubmissions by poisoning the interrupt URBs at disconnect and using a
dedicated disconnected flag.
This also makes sure that all I/O has completed by the time the
disconnect callback returns.
Fixes: 1ef37c6047 ("USB: adutux: remove custom debug macro and module parameter")
Fixes: 66d4bc30d1 ("USB: adutux: remove custom debug macro")
Cc: stable <stable@vger.kernel.org> # 3.12
Signed-off-by: Johan Hovold <johan@kernel.org>
Link: https://lore.kernel.org/r/20190925092913.8608-2-johan@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 8de66b0e6a upstream.
The system can hit a deadlock if an xhci adapter breaks while initializing.
The deadlock is between two threads: thread 1 is tearing down the
adapter and is stuck in usb_unlocked_disable_lpm waiting to lock the
hcd->handwidth_mutex. Thread 2 is holding this mutex (while still trying
to add a usb device), but is stuck in xhci_endpoint_reset waiting for a
stop or config command to complete. A reboot is required to resolve.
It turns out when calling xhci_queue_stop_endpoint and
xhci_queue_configure_endpoint in xhci_endpoint_reset, the return code is
not checked for errors. If the timing is right and the adapter dies just
before either of these commands get issued, we hang indefinitely waiting
for a completion on a command that didn't get issued.
This wasn't a problem before the following fix because we didn't send
commands in xhci_endpoint_reset:
commit f5249461b5 ("xhci: Clear the host side toggle manually when
endpoint is soft reset")
With the patch I am submitting, a duration test which breaks adapters
during initialization (and which deadlocks with the standard kernel) runs
without issue.
Fixes: f5249461b5 ("xhci: Clear the host side toggle manually when endpoint is soft reset")
Cc: <stable@vger.kernel.org> # v4.17+
Cc: Torez Smith <torez@redhat.com>
Signed-off-by: Bill Kuzeja <william.kuzeja@stratus.com>
Signed-off-by: Torez Smith <torez@redhat.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Link: https://lore.kernel.org/r/1570190373-30684-7-git-send-email-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit c03101ff4f upstream.
The check printing out the "WARN Wrong bounce buffer write length:"
uses incorrect values when comparing bytes written from scatterlist
to bounce buffer. Actual copied lengths are fine.
The used seg->bounce_len will be set to equal new_buf_len a few lines later
in the code, but is incorrect when doing the comparison.
The patch which added this false warning was backported to 4.8+ kernels
so this should be backported as far as well.
Cc: <stable@vger.kernel.org> # v4.8+
Fixes: 597c56e372 ("xhci: update bounce buffer with correct sg num")
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Link: https://lore.kernel.org/r/1570190373-30684-2-git-send-email-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit bed5ef2309 upstream.
The driver was using its struct usb_interface pointer as an inverted
disconnected flag and was setting it to NULL before making sure all
completion handlers had run. This could lead to NULL-pointer
dereferences in the dev_err() statements in the completion handlers
which relies on said pointer.
Fix this by using a dedicated disconnected flag.
Note that this is also addresses a NULL-pointer dereference at release()
and a struct usb_interface reference leak introduced by a recent runtime
PM fix, which depends on and should have been submitted together with
this patch.
Fixes: 4212cd74ca ("USB: usb-skeleton.c: remove err() usage")
Fixes: 5c290a5e42 ("USB: usb-skeleton: fix runtime PM after driver unbind")
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Johan Hovold <johan@kernel.org>
Link: https://lore.kernel.org/r/20191009170944.30057-2-johan@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 5c290a5e42 upstream.
Since commit c2b71462d2 ("USB: core: Fix bug caused by duplicate
interface PM usage counter") USB drivers must always balance their
runtime PM gets and puts, including when the driver has already been
unbound from the interface.
Leaving the interface with a positive PM usage counter would prevent a
later bound driver from suspending the device.
Fixes: c2b71462d2 ("USB: core: Fix bug caused by duplicate interface PM usage counter")
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Johan Hovold <johan@kernel.org>
Link: https://lore.kernel.org/r/20191001084908.2003-2-johan@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit aafb00a977 upstream.
The driver was using its struct usb_interface pointer as an inverted
disconnected flag, but was setting it to NULL without making sure all
code paths that used it were done with it.
Before commit ef61eb43ad ("USB: yurex: Fix protection fault after
device removal") this included the interrupt-in completion handler, but
there are further accesses in dev_err and dev_dbg statements in
yurex_write() and the driver-data destructor (sic!).
Fix this by unconditionally stopping also the control URB at disconnect
and by using a dedicated disconnected flag.
Note that we need to take a reference to the struct usb_interface to
avoid a use-after-free in the destructor whenever the device was
disconnected while the character device was still open.
Fixes: aadd6472d9 ("USB: yurex.c: remove dbg() usage")
Fixes: 45714104b9 ("USB: yurex.c: remove err() usage")
Cc: stable <stable@vger.kernel.org> # 3.5: ef61eb43ad
Signed-off-by: Johan Hovold <johan@kernel.org>
Link: https://lore.kernel.org/r/20191009153848.8664-6-johan@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 32a0721c66 upstream.
According to Greg KH, it has been generally agreed that when a USB
driver encounters an unknown error (or one it can't handle directly),
it should just give up instead of going into a potentially infinite
retry loop.
The three codes -EPROTO, -EILSEQ, and -ETIME fall into this category.
They can be caused by bus errors such as packet loss or corruption,
attempting to communicate with a disconnected device, or by malicious
firmware. Nowadays the extent of packet loss or corruption is
negligible, so it should be safe for a driver to give up whenever one
of these errors occurs.
Although the yurex driver handles -EILSEQ errors in this way, it
doesn't do the same for -EPROTO (as discovered by the syzbot fuzzer)
or other unrecognized errors. This patch adjusts the driver so that
it doesn't log an error message for -EPROTO or -ETIME, and it doesn't
retry after any errors.
Reported-and-tested-by: syzbot+b24d736f18a1541ad550@syzkaller.appspotmail.com
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: Tomoki Sekiyama <tomoki.sekiyama@gmail.com>
CC: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/Pine.LNX.4.44L0.1909171245410.1590-100000@iolanthe.rowland.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 38fb6d0ea3 ]
The kernel mount_block_root() function expects -EACESS or -EINVAL for a
unmountable filesystem when trying to mount the root with different
filesystem types.
However, in 5.3-rc1 the behavior when F2FS code cannot find valid block
changed to return -EFSCORRUPTED(-EUCLEAN), and this error code makes
mount_block_root() fail when trying to probe F2FS.
When the magic number of the superblock mismatches, it has a high
probability that it's just not a F2FS. In this case return -EINVAL seems
to be a better result, and this return value can make mount_block_root()
probing work again.
Return -EINVAL when the superblock has magic mismatch, -EFSCORRUPTED in
other cases (the magic matches but the superblock cannot be recognized).
Fixes: 10f966bbf5 ("f2fs: use generic EFSBADCRC/EFSCORRUPTED")
Signed-off-by: Icenowy Zheng <icenowy@aosc.io>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 20bb759a66 upstream.
Calling 'panic()' on a kernel with CONFIG_PREEMPT=y can leave the
calling CPU in an infinite loop, but with interrupts and preemption
enabled. From this state, userspace can continue to be scheduled,
despite the system being "dead" as far as the kernel is concerned.
This is easily reproducible on arm64 when booting with "nosmp" on the
command line; a couple of shell scripts print out a periodic "Ping"
message whilst another triggers a crash by writing to
/proc/sysrq-trigger:
| sysrq: Trigger a crash
| Kernel panic - not syncing: sysrq triggered crash
| CPU: 0 PID: 1 Comm: init Not tainted 5.2.15 #1
| Hardware name: linux,dummy-virt (DT)
| Call trace:
| dump_backtrace+0x0/0x148
| show_stack+0x14/0x20
| dump_stack+0xa0/0xc4
| panic+0x140/0x32c
| sysrq_handle_reboot+0x0/0x20
| __handle_sysrq+0x124/0x190
| write_sysrq_trigger+0x64/0x88
| proc_reg_write+0x60/0xa8
| __vfs_write+0x18/0x40
| vfs_write+0xa4/0x1b8
| ksys_write+0x64/0xf0
| __arm64_sys_write+0x14/0x20
| el0_svc_common.constprop.0+0xb0/0x168
| el0_svc_handler+0x28/0x78
| el0_svc+0x8/0xc
| Kernel Offset: disabled
| CPU features: 0x0002,24002004
| Memory Limit: none
| ---[ end Kernel panic - not syncing: sysrq triggered crash ]---
| Ping 2!
| Ping 1!
| Ping 1!
| Ping 2!
The issue can also be triggered on x86 kernels if CONFIG_SMP=n,
otherwise local interrupts are disabled in 'smp_send_stop()'.
Disable preemption in 'panic()' before re-enabling interrupts.
Link: http://lkml.kernel.org/r/20191002123538.22609-1-will@kernel.org
Link: https://lore.kernel.org/r/BX1W47JXPMR8.58IYW53H6M5N@dragonstone
Signed-off-by: Will Deacon <will@kernel.org>
Reported-by: Xogium <contact@xogium.me>
Reviewed-by: Kees Cook <keescook@chromium.org>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Petr Mladek <pmladek@suse.com>
Cc: Feng Tang <feng.tang@intel.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 7388afe091 upstream.
Enforce the first argument to be a correct type of a pointer to struct
element and avoid unnecessary typecasts from const to non-const pointers
(the change in validate_ie_attr() is needed to make this part work). In
addition, avoid signed/unsigned comparison within for_each_element() and
mark struct element packed just in case.
Signed-off-by: Jouni Malinen <j@w1.fi>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 0f3b07f027 upstream.
Rather than always iterating elements from frames with pure
u8 pointers, add a type "struct element" that encapsulates
the id/datalen/data format of them.
Then, add the element iteration macros
* for_each_element
* for_each_element_id
* for_each_element_extid
which take, as their first 'argument', such a structure and
iterate through a given u8 array interpreting it as elements.
While at it and since we'll need it, also add
* for_each_subelement
* for_each_subelement_id
* for_each_subelement_extid
which instead of taking data/length just take an outer element
and use its data/datalen.
Also add for_each_element_completed() to determine if any of
the loops above completed, i.e. it was able to parse all of
the elements successfully and no data remained.
Use for_each_element_id() in cfg80211_find_ie_match() as the
first user of this.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit ee45197c80 upstream.
As reported by erofs_utils fuzzer, a logical page can belong
to at most 2 compressed clusters, if one compressed cluster
is corrupted, but the other has been ready in submitting chain.
The chain needs to submit anyway in order to keep the page
working properly (page unlocked with PG_error set, PG_uptodate
not set).
Let's fix it now.
Fixes: 3883a79abd ("staging: erofs: introduce VLE decompression support")
Cc: <stable@vger.kernel.org> # 4.19+
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Link: https://lore.kernel.org/r/20190819103426.87579-2-gaoxiang25@huawei.com
[ Gao Xiang: Manually backport to v4.19.y stable. ]
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 1004ce4c25 upstream.
Synchronization is recommended before disabling the trace registers
to prevent any start or stop points being speculative at the point
of disabling the unit (section 7.3.77 of ARM IHI 0064D).
Synchronization is also recommended after programming the trace
registers to ensure all updates are committed prior to normal code
resuming (section 4.3.7 of ARM IHI 0064D).
Let's ensure these syncronization points are present in the code
and clearly commented.
Note that we could rely on the barriers in CS_LOCK and
coresight_disclaim_device_unlocked or the context switch to user
space - however coresight may be of use in the kernel.
On armv8 the mb macro is defined as dsb(sy) - Given that the etm4x is
only used on armv8 let's directly use dsb(sy) instead of mb(). This
removes some ambiguity and makes it easier to correlate the code with
the TRM.
Signed-off-by: Andrew Murray <andrew.murray@arm.com>
Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
[Fixed capital letter for "use" in title]
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Link: https://lore.kernel.org/r/20190829202842.580-11-mathieu.poirier@linaro.org
Cc: stable@vger.kernel.org # 4.9+
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit cc3a7bfe62 upstream.
Today, put_compat_statfs64() disallows nearly any field value over
2^32 if f_bsize is only 32 bits, but that makes no sense.
compat_statfs64 is there for the explicit purpose of providing 64-bit
fields for f_files, f_ffree, etc. And f_bsize is always only 32 bits.
As a result, 32-bit userspace gets -EOVERFLOW for i.e. large file
counts even with -D_FILE_OFFSET_BITS=64 set.
In reality, only f_bsize and f_frsize can legitimately overflow
(fields like f_type and f_namelen should never be large), so test
only those fields.
This bug was discussed at length some time ago, and this is the proposal
Al suggested at https://lkml.org/lkml/2018/8/6/640. It seemed to get
dropped amid the discussion of other related changes, but this
part seems obviously correct on its own, so I've picked it up and
sent it, for expediency.
Fixes: 64d2ab32ef ("vfs: fix put_compat_statfs64() does not handle errors")
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit a111b7c0f2 upstream.
Configure arm64 runtime CPU speculation bug mitigations in accordance
with the 'mitigations=' cmdline option. This affects Meltdown, Spectre
v2, and Speculative Store Bypass.
The default behavior is unchanged.
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
[will: reorder checks so KASLR implies KPTI and SSBS is affected by cmdline]
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit cbdf8a189a ]
On a CPU that doesn't support SSBS, PSTATE[12] is RES0. In a system
where only some of the CPUs implement SSBS, we end-up losing track of
the SSBS bit across task migration.
To address this issue, let's force the SSBS bit on context switch.
Fixes: 8f04e8e6e2 ("arm64: ssbd: Add support for PSTATE.SSBS rather than trapping to EL3")
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
[will: inverted logic and added comments]
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit eb337cdfcd ]
SSBS provides a relatively cheap mitigation for SSB, but it is still a
mitigation and its presence does not indicate that the CPU is unaffected
by the vulnerability.
Tweak the mitigation logic so that we report the correct string in sysfs.
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 526e065dbc ]
Return status based on ssbd_state and __ssb_safe. If the
mitigation is disabled, or the firmware isn't responding then
return the expected machine state based on a whitelist of known
good cores.
Given a heterogeneous machine, the overall machine vulnerability
defaults to safe but is reset to unsafe when we miss the whitelist
and the firmware doesn't explicitly tell us the core is safe.
In order to make that work we delay transitioning to vulnerable
until we know the firmware isn't responding to avoid a case
where we miss the whitelist, but the firmware goes ahead and
reports the core is not vulnerable. If all the cores in the
machine have SSBS, then __ssb_safe will remain true.
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Jeremy Linton <jeremy.linton@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 73f3816609 ]
We currently have a list of CPUs affected by Spectre-v2, for which
we check that the firmware implements ARCH_WORKAROUND_1. It turns
out that not all firmwares do implement the required mitigation,
and that we fail to let the user know about it.
Instead, let's slightly revamp our checks, and rely on a whitelist
of cores that are known to be non-vulnerable, and let the user know
the status of the mitigation in the kernel log.
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Jeremy Linton <jeremy.linton@arm.com>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit f54dada827 ]
In valid_user_regs() we treat SSBS as a RES0 bit, and consequently it is
unexpectedly cleared when we restore a sigframe or fiddle with GPRs via
ptrace.
This patch fixes valid_user_regs() to account for this, updating the
function to refer to the latest ARM ARM (ARM DDI 0487D.a). For AArch32
tasks, SSBS appears in bit 23 of SPSR_EL1, matching its position in the
AArch32-native PSR format, and we don't need to translate it as we have
to for DIT.
There are no other bit assignments that we need to account for today.
As the recent documentation describes the DIT bit, we can drop our
comment regarding DIT.
While removing SSBS from the RES0 masks, existing inconsistent
whitespace is corrected.
Fixes: d71be2b6c0 ("arm64: cpufeature: Detect SSBS and advertise to userspace")
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Suzuki K Poulose <suzuki.poulose@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 8f04e8e6e2 ]
On CPUs with support for PSTATE.SSBS, the kernel can toggle the SSBD
state without needing to call into firmware.
This patch hooks into the existing SSBD infrastructure so that SSBS is
used on CPUs that support it, but it's all made horribly complicated by
the very real possibility of big/little systems that don't uniformly
provide the new capability.
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit c82dd6d078 ]
When the handle_exception function addresses an exception, the interrupts
will be unconditionally enabled after finishing the context save. However,
It may erroneously enable the interrupts if the interrupts are disabled
before entering the handle_exception.
For example, one of the WARN_ON() condition is satisfied in the scheduling
where the interrupt is disabled and rq.lock is locked. The WARN_ON will
trigger a break exception and the handle_exception function will enable the
interrupts before entering do_trap_break function. During the procedure, if
a timer interrupt is pending, it will be taken when interrupts are enabled.
In this case, it may cause a deadlock problem if the rq.lock is locked
again in the timer ISR.
Hence, the handle_exception() can only enable interrupts when the state of
sstatus.SPIE is 1.
This patch is tested on HiFive Unleashed board.
Signed-off-by: Vincent Chen <vincent.chen@sifive.com>
Reviewed-by: Palmer Dabbelt <palmer@sifive.com>
[paul.walmsley@sifive.com: updated to apply]
Fixes: bcae803a21 ("RISC-V: Enable IRQ during exception handling")
Cc: David Abdurachmanov <david.abdurachmanov@sifive.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paul Walmsley <paul.walmsley@sifive.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit b63fd11cce ]
When using 'perf stat' with repeat and interval option, it shows wrong
values for events.
The wrong values will be shown for the first interval on the second and
subsequent repetitions.
Without the fix:
# perf stat -r 3 -I 2000 -e faults -e sched:sched_switch -a sleep 5
2.000282489 53 faults
2.000282489 513 sched:sched_switch
4.005478208 3,721 faults
4.005478208 2,666 sched:sched_switch
5.025470933 395 faults
5.025470933 1,307 sched:sched_switch
2.009602825 1,84,46,74,40,73,70,95,47,520 faults <------
2.009602825 1,84,46,74,40,73,70,95,49,568 sched:sched_switch <------
4.019612206 4,730 faults
4.019612206 2,746 sched:sched_switch
5.039615484 3,953 faults
5.039615484 1,496 sched:sched_switch
2.000274620 1,84,46,74,40,73,70,95,47,520 faults <------
2.000274620 1,84,46,74,40,73,70,95,47,520 sched:sched_switch <------
4.000480342 4,282 faults
4.000480342 2,303 sched:sched_switch
5.000916811 1,322 faults
5.000916811 1,064 sched:sched_switch
#
prev_raw_counts is allocated when using intervals. This is used when
calculating the difference in the counts of events when using interval.
The current counts are stored in prev_raw_counts to calculate the
differences in the next iteration.
On the first interval of the second and subsequent repetitions,
prev_raw_counts would be the values stored in the last interval of the
previous repetitions, while the current counts will only be for the
first interval of the current repetition.
Hence there is a possibility of events showing up as big number.
Fix this by resetting prev_raw_counts whenever perf stat repeats the
command.
With the fix:
# perf stat -r 3 -I 2000 -e faults -e sched:sched_switch -a sleep 5
2.019349347 2,597 faults
2.019349347 2,753 sched:sched_switch
4.019577372 3,098 faults
4.019577372 2,532 sched:sched_switch
5.019415481 1,879 faults
5.019415481 1,356 sched:sched_switch
2.000178813 8,468 faults
2.000178813 2,254 sched:sched_switch
4.000404621 7,440 faults
4.000404621 1,266 sched:sched_switch
5.040196079 2,458 faults
5.040196079 556 sched:sched_switch
2.000191939 6,870 faults
2.000191939 1,170 sched:sched_switch
4.000414103 541 faults
4.000414103 902 sched:sched_switch
5.000809863 450 faults
5.000809863 364 sched:sched_switch
#
Committer notes:
This was broken since the cset introducing the --interval feature, i.e.
--repeat + --interval wasn't tested at that point, add the Fixes tag so
that automatic scripts can pick this up.
Fixes: 13370a9b5b ("perf stat: Add interval printing")
Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Tested-by: Ravi Bangoria <ravi.bangoria@linux.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: stable@vger.kernel.org # v3.9+
Link: http://lore.kernel.org/lkml/20190904094738.9558-2-srikar@linux.vnet.ibm.com
[ Fixed up conflicts with libperf, i.e. some perf_{evsel,evlist} lost the 'perf' prefix ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 0216234c2e ]
We release wrong pointer on error path in cpu_cache_level__read
function, leading to segfault:
(gdb) r record ls
Starting program: /root/perf/tools/perf/perf record ls
...
[ perf record: Woken up 1 times to write data ]
double free or corruption (out)
Thread 1 "perf" received signal SIGABRT, Aborted.
0x00007ffff7463798 in raise () from /lib64/power9/libc.so.6
(gdb) bt
#0 0x00007ffff7463798 in raise () from /lib64/power9/libc.so.6
#1 0x00007ffff7443bac in abort () from /lib64/power9/libc.so.6
#2 0x00007ffff74af8bc in __libc_message () from /lib64/power9/libc.so.6
#3 0x00007ffff74b92b8 in malloc_printerr () from /lib64/power9/libc.so.6
#4 0x00007ffff74bb874 in _int_free () from /lib64/power9/libc.so.6
#5 0x0000000010271260 in __zfree (ptr=0x7fffffffa0b0) at ../../lib/zalloc..
#6 0x0000000010139340 in cpu_cache_level__read (cache=0x7fffffffa090, cac..
#7 0x0000000010143c90 in build_caches (cntp=0x7fffffffa118, size=<optimiz..
...
Releasing the proper pointer.
Fixes: 720e98b5fa ("perf tools: Add perf data cache feature")
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: stable@vger.kernel.org: # v4.6+
Link: http://lore.kernel.org/lkml/20190912105235.10689-1-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit b9023b91dd ]
When a cpu requests broadcasting, before starting the tick broadcast
hrtimer, bc_set_next() checks if the timer callback (bc_handler) is active
using hrtimer_try_to_cancel(). But hrtimer_try_to_cancel() does not provide
the required synchronization when the callback is active on other core.
The callback could have already executed tick_handle_oneshot_broadcast()
and could have also returned. But still there is a small time window where
the hrtimer_try_to_cancel() returns -1. In that case bc_set_next() returns
without doing anything, but the next_event of the tick broadcast clock
device is already set to a timeout value.
In the race condition diagram below, CPU #1 is running the timer callback
and CPU #2 is entering idle state and so calls bc_set_next().
In the worst case, the next_event will contain an expiry time, but the
hrtimer will not be started which happens when the racing callback returns
HRTIMER_NORESTART. The hrtimer might never recover if all further requests
from the CPUs to subscribe to tick broadcast have timeout greater than the
next_event of tick broadcast clock device. This leads to cascading of
failures and finally noticed as rcu stall warnings
Here is a depiction of the race condition
CPU #1 (Running timer callback) CPU #2 (Enter idle
and subscribe to
tick broadcast)
--------------------- ---------------------
__run_hrtimer() tick_broadcast_enter()
bc_handler() __tick_broadcast_oneshot_control()
tick_handle_oneshot_broadcast()
raw_spin_lock(&tick_broadcast_lock);
dev->next_event = KTIME_MAX; //wait for tick_broadcast_lock
//next_event for tick broadcast clock
set to KTIME_MAX since no other cores
subscribed to tick broadcasting
raw_spin_unlock(&tick_broadcast_lock);
if (dev->next_event == KTIME_MAX)
return HRTIMER_NORESTART
// callback function exits without
restarting the hrtimer //tick_broadcast_lock acquired
raw_spin_lock(&tick_broadcast_lock);
tick_broadcast_set_event()
clockevents_program_event()
dev->next_event = expires;
bc_set_next()
hrtimer_try_to_cancel()
//returns -1 since the timer
callback is active. Exits without
restarting the timer
cpu_base->running = NULL;
The comment that hrtimer cannot be armed from within the callback is
wrong. It is fine to start the hrtimer from within the callback. Also it is
safe to start the hrtimer from the enter/exit idle code while the broadcast
handler is active. The enter/exit idle code and the broadcast handler are
synchronized using tick_broadcast_lock. So there is no need for the
existing try to cancel logic. All this can be removed which will eliminate
the race condition as well.
Fixes: 5d1638acb9 ("tick: Introduce hrtimer based broadcast")
Originally-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Balasubramani Vivekanandan <balasubramani_vivekanandan@mentor.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org
Link: https://lkml.kernel.org/r/20190926135101.12102-2-balasubramani_vivekanandan@mentor.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
