One of the things that CONFIG_HARDENED_USERCOPY sanity-checks is whether
an object that is about to be copied to/from userspace is overlapping
the stack at all. If it is, it performs a number of inexpensive
bounds checks. One of the finer-grained checks is whether an object
crosses stack frames within the stack region. Doing this on x86 with
CONFIG_FRAME_POINTER was cheap/easy. Doing it with ORC was deemed too
heavy, and was left out (a while ago), leaving the courser whole-stack
check.
The LKDTM tests USERCOPY_STACK_FRAME_TO and USERCOPY_STACK_FRAME_FROM
try to exercise these cross-frame cases to validate the defense is
working. They have been failing ever since ORC was added (which was
expected). While Muhammad was investigating various LKDTM failures[1],
he asked me for additional details on them, and I realized that when
exact stack frame boundary checking is not available (i.e. everything
except x86 with FRAME_POINTER), it could check if a stack object is at
least "current depth valid", in the sense that any object within the
stack region but not between start-of-stack and current_stack_pointer
should be considered unavailable (i.e. its lifetime is from a call no
longer present on the stack).
Introduce ARCH_HAS_CURRENT_STACK_POINTER to track which architectures
have actually implemented the common global register alias.
Additionally report usercopy bounds checking failures with an offset
from current_stack_pointer, which may assist with diagnosing failures.
The LKDTM USERCOPY_STACK_FRAME_TO and USERCOPY_STACK_FRAME_FROM tests
(once slightly adjusted in a separate patch) pass again with this fixed.
[1] https://github.com/kernelci/kernelci-project/issues/84
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: linux-mm@kvack.org
Reported-by: Muhammad Usama Anjum <usama.anjum@collabora.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
---
v1: https://lore.kernel.org/lkml/20220216201449.2087956-1-keescook@chromium.org
v2: https://lore.kernel.org/lkml/20220224060342.1855457-1-keescook@chromium.org
v3: https://lore.kernel.org/lkml/20220225173345.3358109-1-keescook@chromium.org
v4: - improve commit log (akpm)
All supported versions of Clang perform auto-init of __builtin_alloca()
when stack auto-init is on (CONFIG_INIT_STACK_ALL_{ZERO,PATTERN}).
add_random_kstack_offset() uses __builtin_alloca() to add a stack
offset. This means, when CONFIG_INIT_STACK_ALL_{ZERO,PATTERN} is
enabled, add_random_kstack_offset() will auto-init that unused portion
of the stack used to add an offset.
There are several problems with this:
1. These offsets can be as large as 1023 bytes. Performing
memset() on them isn't exactly cheap, and this is done on
every syscall entry.
2. Architectures adding add_random_kstack_offset() to syscall
entry implemented in C require them to be 'noinstr' (e.g. see
x86 and s390). The potential problem here is that a call to
memset may occur, which is not noinstr.
A x86_64 defconfig kernel with Clang 11 and CONFIG_VMLINUX_VALIDATION shows:
| vmlinux.o: warning: objtool: do_syscall_64()+0x9d: call to memset() leaves .noinstr.text section
| vmlinux.o: warning: objtool: do_int80_syscall_32()+0xab: call to memset() leaves .noinstr.text section
| vmlinux.o: warning: objtool: __do_fast_syscall_32()+0xe2: call to memset() leaves .noinstr.text section
| vmlinux.o: warning: objtool: fixup_bad_iret()+0x2f: call to memset() leaves .noinstr.text section
Clang 14 (unreleased) will introduce a way to skip alloca initialization
via __builtin_alloca_uninitialized() (https://reviews.llvm.org/D115440).
Constrain RANDOMIZE_KSTACK_OFFSET to only be enabled if no stack
auto-init is enabled, the compiler is GCC, or Clang is version 14+. Use
__builtin_alloca_uninitialized() if the compiler provides it, as is done
by Clang 14.
Link: https://lkml.kernel.org/r/YbHTKUjEejZCLyhX@elver.google.com
Fixes: 39218ff4c6 ("stack: Optionally randomize kernel stack offset each syscall")
Signed-off-by: Marco Elver <elver@google.com>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Acked-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/r/20220131090521.1947110-2-elver@google.com
The randomize_kstack_offset feature is unconditionally compiled in when
the architecture supports it.
To add constraints on compiler versions, we require a dedicated Kconfig
variable. Therefore, introduce RANDOMIZE_KSTACK_OFFSET.
Furthermore, this option is now also configurable by EXPERT kernels:
while the feature is supposed to have zero performance overhead when
disabled, due to its use of static branches, there are few cases where
giving a distribution the option to disable the feature entirely makes
sense. For example, in very resource constrained environments, which
would never enable the feature to begin with, in which case the
additional kernel code size increase would be redundant.
Signed-off-by: Marco Elver <elver@google.com>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/r/20220131090521.1947110-1-elver@google.com
The .noinstr.text section functions may not have "current()" sanely
available. Similarly true for .entry.text, though such a check is
currently redundant. Add a check for both. In an x86_64 defconfig build,
the following functions no longer receive stackleak instrumentation:
__do_fast_syscall_32()
do_int80_syscall_32()
do_machine_check()
do_syscall_64()
exc_general_protection()
fixup_bad_iret()
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Cc: Alexander Popov <alex.popov@linux.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Since STRING_CST may not be NUL terminated, strncmp() was used for check
for equality. However, this may lead to mismatches for longer section
names where the start matches the tested-for string. Test for exact
equality by checking for the presences of NUL termination.
Cc: Alexander Popov <alex.popov@linux.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
In order to compare instrumentation between builds, make the verbose
mode of the plugin available during the build. This is rarely needed
(behind EXPERT) and very noisy (disabled for COMPILE_TEST).
Cc: Alexander Popov <alex.popov@linux.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Pull irq fixes from Borislav Petkov:
- Drop an unused private data field in the AIC driver
- Various fixes to the realtek-rtl driver
- Make the GICv3 ITS driver compile again in !SMP configurations
- Force reset of the GICv3 ITSs at probe time to avoid issues during kexec
- Yet another kfree/bitmap_free conversion
- Various DT updates (Renesas, SiFive)
* tag 'irq_urgent_for_v5.17_rc2_p2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
dt-bindings: interrupt-controller: sifive,plic: Group interrupt tuples
dt-bindings: interrupt-controller: sifive,plic: Fix number of interrupts
dt-bindings: irqchip: renesas-irqc: Add R-Car V3U support
irqchip/gic-v3-its: Reset each ITS's BASERn register before probe
irqchip/gic-v3-its: Fix build for !SMP
irqchip/loongson-pch-ms: Use bitmap_free() to free bitmap
irqchip/realtek-rtl: Service all pending interrupts
irqchip/realtek-rtl: Fix off-by-one in routing
irqchip/realtek-rtl: Map control data to virq
irqchip/apple-aic: Drop unused ipi_hwirq field
Pull perf fixes from Borislav Petkov:
- Prevent accesses to the per-CPU cgroup context list from another CPU
except the one it belongs to, to avoid list corruption
- Make sure parent events are always woken up to avoid indefinite hangs
in the traced workload
* tag 'perf_urgent_for_v5.17_rc2_p2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
perf/core: Fix cgroup event list management
perf: Always wake the parent event
Pull scheduler fix from Borislav Petkov:
"Make sure the membarrier-rseq fence commands are part of the reported
set when querying membarrier(2) commands through MEMBARRIER_CMD_QUERY"
* tag 'sched_urgent_for_v5.17_rc2_p2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
sched/membarrier: Fix membarrier-rseq fence command missing from query bitmask
Pull x86 fixes from Borislav Petkov:
- Add another Intel CPU model to the list of CPUs supporting the
processor inventory unique number
- Allow writing to MCE thresholding sysfs files again - a previous
change had accidentally disabled it and no one noticed. Goes to show
how much is this stuff used
* tag 'x86_urgent_for_v5.17_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/cpu: Add Xeon Icelake-D to list of CPUs that support PPIN
x86/MCE/AMD: Allow thresholding interface updates after init
Merge misc fixes from Andrew Morton:
"12 patches.
Subsystems affected by this patch series: sysctl, binfmt, ia64, mm
(memory-failure, folios, kasan, and psi), selftests, and ocfs2"
* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
ocfs2: fix a deadlock when commit trans
jbd2: export jbd2_journal_[grab|put]_journal_head
psi: fix "defined but not used" warnings when CONFIG_PROC_FS=n
psi: fix "no previous prototype" warnings when CONFIG_CGROUPS=n
mm, kasan: use compare-exchange operation to set KASAN page tag
kasan: test: fix compatibility with FORTIFY_SOURCE
tools/testing/scatterlist: add missing defines
mm: page->mapping folio->mapping should have the same offset
memory-failure: fetch compound_head after pgmap_pfn_valid()
ia64: make IA64_MCA_RECOVERY bool instead of tristate
binfmt_misc: fix crash when load/unload module
include/linux/sysctl.h: fix register_sysctl_mount_point() return type
When CONFIG_PROC_FS is disabled psi code generates the following
warnings:
kernel/sched/psi.c:1364:30: warning: 'psi_cpu_proc_ops' defined but not used [-Wunused-const-variable=]
1364 | static const struct proc_ops psi_cpu_proc_ops = {
| ^~~~~~~~~~~~~~~~
kernel/sched/psi.c:1355:30: warning: 'psi_memory_proc_ops' defined but not used [-Wunused-const-variable=]
1355 | static const struct proc_ops psi_memory_proc_ops = {
| ^~~~~~~~~~~~~~~~~~~
kernel/sched/psi.c:1346:30: warning: 'psi_io_proc_ops' defined but not used [-Wunused-const-variable=]
1346 | static const struct proc_ops psi_io_proc_ops = {
| ^~~~~~~~~~~~~~~
Make definitions of these structures and related functions conditional
on CONFIG_PROC_FS config.
Link: https://lkml.kernel.org/r/20220119223940.787748-3-surenb@google.com
Fixes: 0e94682b73 ("psi: introduce psi monitor")
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Reported-by: kernel test robot <lkp@intel.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When CONFIG_CGROUPS is disabled psi code generates the following
warnings:
kernel/sched/psi.c:1112:21: warning: no previous prototype for 'psi_trigger_create' [-Wmissing-prototypes]
1112 | struct psi_trigger *psi_trigger_create(struct psi_group *group,
| ^~~~~~~~~~~~~~~~~~
kernel/sched/psi.c:1182:6: warning: no previous prototype for 'psi_trigger_destroy' [-Wmissing-prototypes]
1182 | void psi_trigger_destroy(struct psi_trigger *t)
| ^~~~~~~~~~~~~~~~~~~
kernel/sched/psi.c:1249:10: warning: no previous prototype for 'psi_trigger_poll' [-Wmissing-prototypes]
1249 | __poll_t psi_trigger_poll(void **trigger_ptr,
| ^~~~~~~~~~~~~~~~
Change the declarations of these functions in the header to provide the
prototypes even when they are unused.
Link: https://lkml.kernel.org/r/20220119223940.787748-2-surenb@google.com
Fixes: 0e94682b73 ("psi: introduce psi monitor")
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Reported-by: kernel test robot <lkp@intel.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
With CONFIG_FORTIFY_SOURCE enabled, string functions will also perform
dynamic checks using __builtin_object_size(ptr), which when failed will
panic the kernel.
Because the KASAN test deliberately performs out-of-bounds operations,
the kernel panics with FORTIFY_SOURCE, for example:
| kernel BUG at lib/string_helpers.c:910!
| invalid opcode: 0000 [#1] PREEMPT SMP KASAN PTI
| CPU: 1 PID: 137 Comm: kunit_try_catch Tainted: G B 5.16.0-rc3+ #3
| Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-2 04/01/2014
| RIP: 0010:fortify_panic+0x19/0x1b
| ...
| Call Trace:
| kmalloc_oob_in_memset.cold+0x16/0x16
| ...
Fix it by also hiding `ptr` from the optimizer, which will ensure that
__builtin_object_size() does not return a valid size, preventing
fortified string functions from panicking.
Link: https://lkml.kernel.org/r/20220124160744.1244685-1-elver@google.com
Signed-off-by: Marco Elver <elver@google.com>
Reported-by: Nico Pache <npache@redhat.com>
Reviewed-by: Nico Pache <npache@redhat.com>
Reviewed-by: Andrey Konovalov <andreyknvl@gmail.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Brendan Higgins <brendanhiggins@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The cited commits replaced preemptible with pagefault_disabled and
flush_kernel_dcache_page with flush_dcache_page respectively, hence need
to update the corresponding defines in the test.
scatterlist.c: In function ‘sg_miter_stop’:
scatterlist.c:919:4: warning: implicit declaration of function ‘flush_dcache_page’ [-Wimplicit-function-declaration]
flush_dcache_page(miter->page);
^~~~~~~~~~~~~~~~~
In file included from linux/scatterlist.h:8:0,
from scatterlist.c:9:
scatterlist.c:922:18: warning: implicit declaration of function ‘pagefault_disabled’ [-Wimplicit-function-declaration]
WARN_ON_ONCE(!pagefault_disabled());
^
linux/mm.h:23:25: note: in definition of macro ‘WARN_ON_ONCE’
int __ret_warn_on = !!(condition); \
^~~~~~~~~
Link: https://lkml.kernel.org/r/20220118082105.1737320-1-maorg@nvidia.com
Fixes: 723aca2085 ("mm/scatterlist: replace the !preemptible warning in sg_miter_stop()")
Fixes: 0e84f5dbf8 ("scatterlist: replace flush_kernel_dcache_page with flush_dcache_page")
Signed-off-by: Maor Gottlieb <maorg@nvidia.com>
Tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In linux-next, IA64_MCA_RECOVERY uses the (new) function
make_task_dead(), which is not exported for use by modules. Instead of
exporting it for one user, convert IA64_MCA_RECOVERY to be a bool
Kconfig symbol.
In a config file from "kernel test robot <lkp@intel.com>" for a
different problem, this linker error was exposed when
CONFIG_IA64_MCA_RECOVERY=m.
Fixes this build error:
ERROR: modpost: "make_task_dead" [arch/ia64/kernel/mca_recovery.ko] undefined!
Link: https://lkml.kernel.org/r/20220124213129.29306-1-rdunlap@infradead.org
Fixes: 0e25498f8c ("exit: Add and use make_task_dead.")
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Suggested-by: Christoph Hellwig <hch@infradead.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Tony Luck <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We should unregister the table upon module unload otherwise something
horrible will happen when we load binfmt_misc module again. Also note
that we should keep value returned by register_sysctl_mount_point() and
release it later, otherwise it will leak.
Also, per Christian's comment, to fully restore the old behavior that
won't break userspace the check(binfmt_misc_header) should be
eliminated.
To reproduce:
modprobe binfmt_misc
modprobe -r binfmt_misc
modprobe binfmt_misc
modprobe -r binfmt_misc
modprobe binfmt_misc
resulting in
modprobe: can't load module binfmt_misc (kernel/fs/binfmt_misc.ko): Cannot allocate memory
and an unhappy kernel:
binfmt_misc: Failed to create fs/binfmt_misc sysctl mount point
binfmt_misc: Failed to create fs/binfmt_misc sysctl mount point
BUG: unable to handle page fault for address: fffffbfff8004802
Call Trace:
init_misc_binfmt+0x2d/0x1000 [binfmt_misc]
Link: https://lkml.kernel.org/r/20220124181812.1869535-2-ztong0001@gmail.com
Fixes: 3ba442d533 ("fs: move binfmt_misc sysctl to its own file")
Signed-off-by: Tong Zhang <ztong0001@gmail.com>
Co-developed-by: Christian Brauner<brauner@kernel.org>
Acked-by: Luis Chamberlain <mcgrof@kernel.org>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Iurii Zaikin <yzaikin@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull irqchip fixes from Marc Zyngier:
- Drop an unused private data field in the AIC driver
- Various fixes to the realtek-rtl driver
- Make the GICv3 ITS driver compile again in !SMP configurations
- Force reset of the GICv3 ITSs at probe time to avoid issues during kexec
- Yet another kfree/bitmap_free conversion
- Various DT updates (Renesas, SiFive)
Link: https://lore.kernel.org/r/20220128174217.517041-1-maz@kernel.org
Pull pci fixes from Bjorn Helgaas:
- Fix compilation warnings in new mt7621 driver (Sergio Paracuellos)
- Restore the sysfs "rom" file for VGA shadow ROMs, which was broken
when converting "rom" to be a static attribute (Bjorn Helgaas)
* tag 'pci-v5.17-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
PCI/sysfs: Find shadow ROM before static attribute initialization
PCI: mt7621: Remove unused function pcie_rmw()
PCI: mt7621: Drop of_match_ptr() to avoid unused variable
Pull gpio fixes from Bartosz Golaszewski:
"Two fixes for the gpio-simulator:
- fix a bug with hogs not being set-up in gpio-sim when user-space
sets the chip label to an empty string
- include the gpio-sim documentation in the index"
* tag 'gpio-fixes-for-v5.17-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/brgl/linux:
gpio: sim: add doc file to index file
gpio: sim: check the label length when setting up device properties
Pull char/misc driver fixes from Greg KH:
"Here are two small char/misc driver fixes for 5.17-rc2 that fix some
reported issues. They are:
- fix up a merge issue in the at25.c driver that ended up dropping
some lines in the driver. The removed lines ended being needed, so
this restores it and the driver works again.
- counter core fix where the wrong error was being returned, NULL
should be the correct error for when memory is gone here, like the
kmalloc() core does.
Both of these have been in linux-next this week with no reported
issues"
* tag 'char-misc-5.17-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
counter: fix an IS_ERR() vs NULL bug
eeprom: at25: Restore missing allocation
Pull tty/serial driver fixes from Greg KH:
"Here are some small bug fixes and reverts for reported problems with
the tty core and drivers. They include:
- revert the fifo use for the 8250 console mode. It caused too many
regressions and problems, and had a bug in it as well. This is
being reworked and should show up in a later -rc1 release, but it's
not ready for 5.17
- rpmsg tty race fix
- restore the cyclades.h uapi header file. Turns out a compiler test
suite used it for some unknown reason. Bring it back just for the
parts that are used by the builder test so they continue to build.
No functionality is restored as no one actually has this hardware
anymore, nor is it really tested.
- stm32 driver fixes
- n_gsm flow control fixes
- pl011 driver fix
- rs485 initialization fix
All of these have been in linux-next this week with no reported
problems"
* tag 'tty-5.17-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
kbuild: remove include/linux/cyclades.h from header file check
serial: core: Initialize rs485 RTS polarity already on probe
serial: pl011: Fix incorrect rs485 RTS polarity on set_mctrl
serial: stm32: fix software flow control transfer
serial: stm32: prevent TDR register overwrite when sending x_char
tty: n_gsm: fix SW flow control encoding/handling
serial: 8250: of: Fix mapped region size when using reg-offset property
tty: rpmsg: Fix race condition releasing tty port
tty: Partially revert the removal of the Cyclades public API
tty: Add support for Brainboxes UC cards.
Revert "tty: serial: Use fifo in 8250 console driver"
Pull USB driver fixes from Greg KH:
"Here are some small USB driver fixes for 5.17-rc2 that resolve a
number of reported problems. These include:
- typec driver fixes
- xhci platform driver fixes for suspending
- ulpi core fix
- role.h build fix
- new device ids
- syzbot-reported bugfixes
- gadget driver fixes
- dwc3 driver fixes
- other small fixes
All of these have been in linux-next this week with no reported
issues"
* tag 'usb-5.17-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
usb: cdnsp: Fix segmentation fault in cdns_lost_power function
usb: dwc2: gadget: don't try to disable ep0 in dwc2_hsotg_suspend
usb: gadget: at91_udc: fix incorrect print type
usb: dwc3: xilinx: Fix error handling when getting USB3 PHY
usb: dwc3: xilinx: Skip resets and USB3 register settings for USB2.0 mode
usb: xhci-plat: fix crash when suspend if remote wake enable
usb: common: ulpi: Fix crash in ulpi_match()
usb: gadget: f_sourcesink: Fix isoc transfer for USB_SPEED_SUPER_PLUS
ucsi_ccg: Check DEV_INT bit only when starting CCG4
USB: core: Fix hang in usb_kill_urb by adding memory barriers
usb-storage: Add unusual-devs entry for VL817 USB-SATA bridge
usb: typec: tcpm: Do not disconnect when receiving VSAFE0V
usb: typec: tcpm: Do not disconnect while receiving VBUS off
usb: typec: Don't try to register component master without components
usb: typec: Only attempt to link USB ports if there is fwnode
usb: typec: tcpci: don't touch CC line if it's Vconn source
usb: roles: fix include/linux/usb/role.h compile issue
Pull block fixes from Jens Axboe:
- NVMe pull request
- add the IGNORE_DEV_SUBNQN quirk for Intel P4500/P4600 SSDs (Wu
Zheng)
- remove the unneeded ret variable in nvmf_dev_show (Changcheng
Deng)
- Fix for a hang regression introduced with a patch in the merge
window, where low queue depth devices would not always get woken
correctly (Laibin)
- Small series fixing an IO accounting issue with bio backed dm devices
(Mike, Yu)
* tag 'block-5.17-2022-01-28' of git://git.kernel.dk/linux-block:
dm: properly fix redundant bio-based IO accounting
dm: revert partial fix for redundant bio-based IO accounting
block: add bio_start_io_acct_time() to control start_time
blk-mq: Fix wrong wakeup batch configuration which will cause hang
nvme-fabrics: remove the unneeded ret variable in nvmf_dev_show
nvme-pci: add the IGNORE_DEV_SUBNQN quirk for Intel P4500/P4600 SSDs
blk-mq: fix missing blk_account_io_done() in error path
block: fix memory leak in disk_register_independent_access_ranges
Pull io_uring fixes from Jens Axboe:
"Just two small fixes this time:
- Fix a bug that can lead to node registration taking 1 second, when
it should finish much quicker (Dylan)
- Remove an unused argument from a function (Usama)"
* tag 'io_uring-5.17-2022-01-28' of git://git.kernel.dk/linux-block:
io_uring: remove unused argument from io_rsrc_node_alloc
io_uring: fix bug in slow unregistering of nodes
Pull powerpc fixes from Michael Ellerman:
- Fix VM debug warnings on boot triggered via __set_fixmap().
- Fix a debug warning in the 64-bit Book3S PMU handling code.
- Fix nested guest HFSCR handling with multiple vCPUs on Power9 or
later.
- Fix decrementer storm caused by a recent change, seen with some
configs.
Thanks to Alexey Kardashevskiy, Athira Rajeev, Christophe Leroy,
Fabiano Rosas, Maxime Bizon, Nicholas Piggin, and Sachin Sant.
* tag 'powerpc-5.17-3' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
powerpc/64s/interrupt: Fix decrementer storm
KVM: PPC: Book3S HV Nested: Fix nested HFSCR being clobbered with multiple vCPUs
powerpc/perf: Fix power_pmu_disable to call clear_pmi_irq_pending only if PMI is pending
powerpc/fixmap: Fix VM debug warning on unmap
Pull arm64 fixes from Catalin Marinas:
- Errata workarounds for Cortex-A510: broken hardware dirty bit
management, detection code for the TRBE (tracing) bugs with the
actual fixes going in via the CoreSight tree.
- Cortex-X2 errata handling for TRBE (inheriting the workarounds from
Cortex-A710).
- Fix ex_handler_load_unaligned_zeropad() to use the correct struct
members.
- A couple of kselftest fixes for FPSIMD.
- Silence the vdso "no previous prototype" warning.
- Mark start_backtrace() notrace and NOKPROBE_SYMBOL.
* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
arm64: cpufeature: List early Cortex-A510 parts as having broken dbm
kselftest/arm64: Correct logging of FPSIMD register read via ptrace
kselftest/arm64: Skip VL_INHERIT tests for unsupported vector types
arm64: errata: Add detection for TRBE trace data corruption
arm64: errata: Add detection for TRBE invalid prohibited states
arm64: errata: Add detection for TRBE ignored system register writes
arm64: Add Cortex-A510 CPU part definition
arm64: extable: fix load_unaligned_zeropad() reg indices
arm64: Mark start_backtrace() notrace and NOKPROBE_SYMBOL
arm64: errata: Update ARM64_ERRATUM_[2119858|2224489] with Cortex-X2 ranges
arm64: Add Cortex-X2 CPU part definition
arm64: vdso: Fix "no previous prototype" warning
Pull security sybsystem fix from James Morris:
"Fix NULL pointer crash in LSM via Ceph, from Vivek Goyal"
* tag 'fixes-v5.17-lsm-ceph-null' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/linux-security:
security, lsm: dentry_init_security() Handle multi LSM registration
Pull documentation fixes from Jonathan Corbet:
"A few documentation fixes for 5.17"
* tag 'docs-5.17-3' of git://git.lwn.net/linux:
docs/vm: Fix typo in *harden*
Documentation: arm: marvell: Extend Avanta list
docs: fix typo in Documentation/kernel-hacking/locking.rst
docs: Hook the RTLA documents into the kernel docs build
Record the start_time for a bio but defer the starting block core's IO
accounting until after IO is submitted using bio_start_io_acct_time().
This approach avoids the need to mess around with any of the
individual IO stats in response to a bio_split() that follows bio
submission.
Reported-by: Bud Brown <bubrown@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: stable@vger.kernel.org
Depends-on: e45c47d1f9 ("block: add bio_start_io_acct_time() to control start_time")
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220128155841.39644-4-snitzer@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
bio_start_io_acct_time() interface is like bio_start_io_acct() that
allows start_time to be passed in. This gives drivers the ability to
defer starting accounting until after IO is issued (but possibily not
entirely due to bio splitting).
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Link: https://lore.kernel.org/r/20220128155841.39644-2-snitzer@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Pull SCSI fixes from James Bottomley:
"Sixteen patches, mostly minor fixes and updates; however there are
substantive driver bug fixes in pm8001, bnx2fc, zfcp, myrs and qedf"
* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
scsi: myrs: Fix crash in error case
scsi: 53c700: Remove redundant assignment to pointer SCp
scsi: ufs: Treat link loss as fatal error
scsi: ufs: Use generic error code in ufshcd_set_dev_pwr_mode()
scsi: bfa: Remove useless DMA-32 fallback configuration
scsi: hisi_sas: Remove useless DMA-32 fallback configuration
scsi: 3w-sas: Remove useless DMA-32 fallback configuration
scsi: bnx2fc: Flush destroy_work queue before calling bnx2fc_interface_put()
scsi: zfcp: Fix failed recovery on gone remote port with non-NPIV FCP devices
scsi: pm8001: Fix bogus FW crash for maxcpus=1
scsi: qedf: Change context reset messages to ratelimited
scsi: qedf: Fix refcount issue when LOGO is received during TMF
scsi: qedf: Add stag_work to all the vports
scsi: ufs: ufshcd-pltfrm: Check the return value of devm_kstrdup()
scsi: target: iscsi: Make sure the np under each tpg is unique
scsi: elx: efct: Don't use GFP_KERNEL under spin lock
Pull EFI fixes from Ard Biesheuvel:
- avoid UEFI v2.00+ runtime services on Apple Mac systems, as they have
been reported to cause crashes, and most Macs claim to be EFI v1.10
anyway
- avoid a spurious boot time warning on arm64 systems with 64k pages
* tag 'efi-urgent-for-v5.17-1' of git://git.kernel.org/pub/scm/linux/kernel/git/efi/efi:
efi: runtime: avoid EFIv2 runtime services on Apple x86 machines
efi/libstub: arm64: Fix image check alignment at entry
A ceph user has reported that ceph is crashing with kernel NULL pointer
dereference. Following is the backtrace.
/proc/version: Linux version 5.16.2-arch1-1 (linux@archlinux) (gcc (GCC)
11.1.0, GNU ld (GNU Binutils) 2.36.1) #1 SMP PREEMPT Thu, 20 Jan 2022
16:18:29 +0000
distro / arch: Arch Linux / x86_64
SELinux is not enabled
ceph cluster version: 16.2.7 (dd0603118f56ab514f133c8d2e3adfc983942503)
relevant dmesg output:
[ 30.947129] BUG: kernel NULL pointer dereference, address:
0000000000000000
[ 30.947206] #PF: supervisor read access in kernel mode
[ 30.947258] #PF: error_code(0x0000) - not-present page
[ 30.947310] PGD 0 P4D 0
[ 30.947342] Oops: 0000 [#1] PREEMPT SMP PTI
[ 30.947388] CPU: 5 PID: 778 Comm: touch Not tainted 5.16.2-arch1-1 #1
86fbf2c313cc37a553d65deb81d98e9dcc2a3659
[ 30.947486] Hardware name: Gigabyte Technology Co., Ltd. B365M
DS3H/B365M DS3H, BIOS F5 08/13/2019
[ 30.947569] RIP: 0010:strlen+0x0/0x20
[ 30.947616] Code: b6 07 38 d0 74 16 48 83 c7 01 84 c0 74 05 48 39 f7 75
ec 31 c0 31 d2 89 d6 89 d7 c3 48 89 f8 31 d2 89 d6 89 d7 c3 0
f 1f 40 00 <80> 3f 00 74 12 48 89 f8 48 83 c0 01 80 38 00 75 f7 48 29 f8 31
ff
[ 30.947782] RSP: 0018:ffffa4ed80ffbbb8 EFLAGS: 00010246
[ 30.947836] RAX: 0000000000000000 RBX: ffffa4ed80ffbc60 RCX:
0000000000000000
[ 30.947904] RDX: 0000000000000000 RSI: 0000000000000000 RDI:
0000000000000000
[ 30.947971] RBP: ffff94b0d15c0ae0 R08: 0000000000000000 R09:
0000000000000000
[ 30.948040] R10: 0000000000000000 R11: 0000000000000000 R12:
0000000000000000
[ 30.948106] R13: 0000000000000001 R14: ffffa4ed80ffbc60 R15:
0000000000000000
[ 30.948174] FS: 00007fc7520f0740(0000) GS:ffff94b7ced40000(0000)
knlGS:0000000000000000
[ 30.948252] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 30.948308] CR2: 0000000000000000 CR3: 0000000104a40001 CR4:
00000000003706e0
[ 30.948376] Call Trace:
[ 30.948404] <TASK>
[ 30.948431] ceph_security_init_secctx+0x7b/0x240 [ceph
49f9c4b9bf5be8760f19f1747e26da33920bce4b]
[ 30.948582] ceph_atomic_open+0x51e/0x8a0 [ceph
49f9c4b9bf5be8760f19f1747e26da33920bce4b]
[ 30.948708] ? get_cached_acl+0x4d/0xa0
[ 30.948759] path_openat+0x60d/0x1030
[ 30.948809] do_filp_open+0xa5/0x150
[ 30.948859] do_sys_openat2+0xc4/0x190
[ 30.948904] __x64_sys_openat+0x53/0xa0
[ 30.948948] do_syscall_64+0x5c/0x90
[ 30.948989] ? exc_page_fault+0x72/0x180
[ 30.949034] entry_SYSCALL_64_after_hwframe+0x44/0xae
[ 30.949091] RIP: 0033:0x7fc7521e25bb
[ 30.950849] Code: 25 00 00 41 00 3d 00 00 41 00 74 4b 64 8b 04 25 18 00
00 00 85 c0 75 67 44 89 e2 48 89 ee bf 9c ff ff ff b8 01 01 0
0 00 0f 05 <48> 3d 00 f0 ff ff 0f 87 91 00 00 00 48 8b 54 24 28 64 48 2b 14
25
Core of the problem is that ceph checks for return code from
security_dentry_init_security() and if return code is 0, it assumes
everything is fine and continues to call strlen(name), which crashes.
Typically SELinux LSM returns 0 and sets name to "security.selinux" and
it is not a problem. Or if selinux is not compiled in or disabled, it
returns -EOPNOTSUP and ceph deals with it.
But somehow in this configuration, 0 is being returned and "name" is
not being initialized and that's creating the problem.
Our suspicion is that BPF LSM is registering a hook for
dentry_init_security() and returns hook default of 0.
LSM_HOOK(int, 0, dentry_init_security, struct dentry *dentry,...)
I have not been able to reproduce it just by doing CONFIG_BPF_LSM=y.
Stephen has tested the patch though and confirms it solves the problem
for him.
dentry_init_security() is written in such a way that it expects only one
LSM to register the hook. Atleast that's the expectation with current code.
If another LSM returns a hook and returns default, it will simply return
0 as of now and that will break ceph.
Hence, suggestion is that change semantics of this hook a bit. If there
are no LSMs or no LSM is taking ownership and initializing security context,
then return -EOPNOTSUP. Also allow at max one LSM to initialize security
context. This hook can't deal with multiple LSMs trying to init security
context. This patch implements this new behavior.
Reported-by: Stephen Muth <smuth4@gmail.com>
Tested-by: Stephen Muth <smuth4@gmail.com>
Suggested-by: Casey Schaufler <casey@schaufler-ca.com>
Acked-by: Casey Schaufler <casey@schaufler-ca.com>
Reviewed-by: Serge Hallyn <serge@hallyn.com>
Cc: Jeff Layton <jlayton@kernel.org>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Paul Moore <paul@paul-moore.com>
Cc: <stable@vger.kernel.org> # 5.16.0
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Acked-by: Paul Moore <paul@paul-moore.com>
Acked-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: James Morris <jmorris@namei.org>
Pull power management fixes from Rafael Wysocki:
"These make the buffer handling in pm_show_wakelocks() more robust and
drop an unused hibernation-related function.
Specifics:
- Make the buffer handling in pm_show_wakelocks() more robust by
using sysfs_emit_at() in it to generate output (Greg
Kroah-Hartman).
- Drop register_nosave_region_late() which is not used (Amadeusz
Sławiński)"
* tag 'pm-5.17-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
PM: hibernate: Remove register_nosave_region_late()
PM: wakeup: simplify the output logic of pm_show_wakelocks()
Pulltracing fixes from Steven Rostedt:
- Limit mcount build time sorting to only those archs that we know it
works for.
- Fix memory leak in error path of histogram setup
- Fix and clean up rel_loc array out of bounds issue
- tools/rtla documentation fixes
- Fix issues with histogram logic
* tag 'trace-v5.17-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace:
tracing: Don't inc err_log entry count if entry allocation fails
tracing: Propagate is_signed to expression
tracing: Fix smatch warning for do while check in event_hist_trigger_parse()
tracing: Fix smatch warning for null glob in event_hist_trigger_parse()
tools/tracing: Update Makefile to build rtla
rtla: Make doc build optional
tracing/perf: Avoid -Warray-bounds warning for __rel_loc macro
tracing: Avoid -Warray-bounds warning for __rel_loc macro
tracing/histogram: Fix a potential memory leak for kstrdup()
ftrace: Have architectures opt-in for mcount build time sorting
