Stephen Boyd [Thu, 8 Jul 2021 01:09:38 +0000 (18:09 -0700)]
scripts/decode_stacktrace.sh: indicate 'auto' can be used for base path
Add "auto" to the usage message so that it's a little clearer that you can
pass "auto" as the second argument. When passing "auto" the script tries
to find the base path automatically instead of requiring it be passed on
the commandline. Also use [<variable>] to indicate the variable argument
and that it is optional so that we can differentiate from the literal
"auto" that should be passed.
Stephen Boyd [Thu, 8 Jul 2021 01:09:35 +0000 (18:09 -0700)]
scripts/decode_stacktrace.sh: silence stderr messages from addr2line/nm
Sometimes if you're using tools that have linked things improperly or have
new features/sections that older tools don't expect you'll see warnings
printed to stderr. We don't really care about these warnings, so let's
just silence these messages to cleanup output of this script.
Stephen Boyd [Thu, 8 Jul 2021 01:09:31 +0000 (18:09 -0700)]
scripts/decode_stacktrace.sh: support debuginfod
Now that stacktraces contain the build ID information we can update this
script to use debuginfod-find to locate the debuginfo for the vmlinux and
modules automatically. This can replace the existing code that requires
specifying a path to vmlinux or tries to find the vmlinux and modules
automatically by using the release number. Work it into the script as a
fallback option if the vmlinux isn't specified on the commandline.
Stephen Boyd [Thu, 8 Jul 2021 01:09:27 +0000 (18:09 -0700)]
x86/dumpstack: use %pSb/%pBb for backtrace printing
Let's use the new printk formats to print the stacktrace entries when
printing a backtrace to the kernel logs. This will include any module's
build ID[1] in it so that offline/crash debugging can easily locate the
debuginfo for a module via something like debuginfod[2].
Stephen Boyd [Thu, 8 Jul 2021 01:09:24 +0000 (18:09 -0700)]
arm64: stacktrace: use %pSb for backtrace printing
Let's use the new printk format to print the stacktrace entry when
printing a backtrace to the kernel logs. This will include any module's
build ID[1] in it so that offline/crash debugging can easily locate the
debuginfo for a module via something like debuginfod[2].
Stephen Boyd [Thu, 8 Jul 2021 01:09:20 +0000 (18:09 -0700)]
module: add printk formats to add module build ID to stacktraces
Let's make kernel stacktraces easier to identify by including the build
ID[1] of a module if the stacktrace is printing a symbol from a module.
This makes it simpler for developers to locate a kernel module's full
debuginfo for a particular stacktrace. Combined with
scripts/decode_stracktrace.sh, a developer can download the matching
debuginfo from a debuginfod[2] server and find the exact file and line
number for the functions plus offsets in a stacktrace that match the
module. This is especially useful for pstore crash debugging where the
kernel crashes are recorded in something like console-ramoops and the
recovery kernel/modules are different or the debuginfo doesn't exist on
the device due to space concerns (the debuginfo can be too large for space
limited devices).
Originally, I put this on the %pS format, but that was quickly rejected
given that %pS is used in other places such as ftrace where build IDs
aren't meaningful. There was some discussions on the list to put every
module build ID into the "Modules linked in:" section of the stacktrace
message but that quickly becomes very hard to read once you have more than
three or four modules linked in. It also provides too much information
when we don't expect each module to be traversed in a stacktrace. Having
the build ID for modules that aren't important just makes things messy.
Splitting it to multiple lines for each module quickly explodes the number
of lines printed in an oops too, possibly wrapping the warning off the
console. And finally, trying to stash away each module used in a
callstack to provide the ID of each symbol printed is cumbersome and would
require changes to each architecture to stash away modules and return
their build IDs once unwinding has completed.
Instead, we opt for the simpler approach of introducing new printk formats
'%pS[R]b' for "pointer symbolic backtrace with module build ID" and '%pBb'
for "pointer backtrace with module build ID" and then updating the few
places in the architecture layer where the stacktrace is printed to use
this new format.
Stephen Boyd [Thu, 8 Jul 2021 01:09:17 +0000 (18:09 -0700)]
dump_stack: add vmlinux build ID to stack traces
Add the running kernel's build ID[1] to the stacktrace information header.
This makes it simpler for developers to locate the vmlinux with full
debuginfo for a particular kernel stacktrace. Combined with
scripts/decode_stracktrace.sh, a developer can download the correct
vmlinux from a debuginfod[2] server and find the exact file and line
number for the functions plus offsets in a stacktrace.
This is especially useful for pstore crash debugging where the kernel
crashes are recorded in the pstore logs and the recovery kernel is
different or the debuginfo doesn't exist on the device due to space
concerns (the data can be large and a security concern). The stacktrace
can be analyzed after the crash by using the build ID to find the matching
vmlinux and understand where in the function something went wrong.
The hex string aa23f7a1231c229de205662d5a9e0d4c580f19a1 is the build ID,
following the kernel version number. Put it all behind a config option,
STACKTRACE_BUILD_ID, so that kernel developers can remove this
information if they decide it is too much.
Stephen Boyd [Thu, 8 Jul 2021 01:09:13 +0000 (18:09 -0700)]
buildid: stash away kernels build ID on init
Parse the kernel's build ID at initialization so that other code can print
a hex format string representation of the running kernel's build ID. This
will be used in the kdump and dump_stack code so that developers can
easily locate the vmlinux debug symbols for a crash/stacktrace.
Stephen Boyd [Thu, 8 Jul 2021 01:09:06 +0000 (18:09 -0700)]
buildid: only consider GNU notes for build ID parsing
Patch series "Add build ID to stacktraces", v6.
This series adds the kernel's build ID[1] to the stacktrace header printed
in oops messages, warnings, etc. and the build ID for any module that
appears in the stacktrace after the module name. The goal is to make the
stacktrace more self-contained and descriptive by including the relevant
build IDs in the kernel logs when something goes wrong. This can be used
by post processing tools like script/decode_stacktrace.sh and kernel
developers to easily locate the debug info associated with a kernel crash
and line up what line and file things started falling apart at.
To show how this can be used I've included a patch to decode_stacktrace.sh
that downloads the debuginfo from a debuginfod server. This also includes
some patches to make the buildid.c file use more const arguments and
consolidate logic into buildid.c from kdump. These are left to the end as
they were mostly cleanup patches.
Some kernel elf files have various notes that also happen to have an elf
note type of '3', which matches NT_GNU_BUILD_ID but the note name isn't
"GNU". For example, this note trips up the existing logic:
Mike Rapoport [Thu, 8 Jul 2021 01:08:15 +0000 (18:08 -0700)]
secretmem: test: add basic selftest for memfd_secret(2)
The test verifies that file descriptor created with memfd_secret does not
allow read/write operations, that secret memory mappings respect
RLIMIT_MEMLOCK and that remote accesses with process_vm_read() and
ptrace() to the secret memory fail.
Mike Rapoport [Thu, 8 Jul 2021 01:08:07 +0000 (18:08 -0700)]
PM: hibernate: disable when there are active secretmem users
It is unsafe to allow saving of secretmem areas to the hibernation
snapshot as they would be visible after the resume and this essentially
will defeat the purpose of secret memory mappings.
Prevent hibernation whenever there are active secret memory users.
Mike Rapoport [Thu, 8 Jul 2021 01:08:03 +0000 (18:08 -0700)]
mm: introduce memfd_secret system call to create "secret" memory areas
Introduce "memfd_secret" system call with the ability to create memory
areas visible only in the context of the owning process and not mapped not
only to other processes but in the kernel page tables as well.
The secretmem feature is off by default and the user must explicitly
enable it at the boot time.
Once secretmem is enabled, the user will be able to create a file
descriptor using the memfd_secret() system call. The memory areas created
by mmap() calls from this file descriptor will be unmapped from the kernel
direct map and they will be only mapped in the page table of the processes
that have access to the file descriptor.
Secretmem is designed to provide the following protections:
* Enhanced protection (in conjunction with all the other in-kernel
attack prevention systems) against ROP attacks. Seceretmem makes
"simple" ROP insufficient to perform exfiltration, which increases the
required complexity of the attack. Along with other protections like
the kernel stack size limit and address space layout randomization which
make finding gadgets is really hard, absence of any in-kernel primitive
for accessing secret memory means the one gadget ROP attack can't work.
Since the only way to access secret memory is to reconstruct the missing
mapping entry, the attacker has to recover the physical page and insert
a PTE pointing to it in the kernel and then retrieve the contents. That
takes at least three gadgets which is a level of difficulty beyond most
standard attacks.
* Prevent cross-process secret userspace memory exposures. Once the
secret memory is allocated, the user can't accidentally pass it into the
kernel to be transmitted somewhere. The secreremem pages cannot be
accessed via the direct map and they are disallowed in GUP.
* Harden against exploited kernel flaws. In order to access secretmem,
a kernel-side attack would need to either walk the page tables and
create new ones, or spawn a new privileged uiserspace process to perform
secrets exfiltration using ptrace.
The file descriptor based memory has several advantages over the
"traditional" mm interfaces, such as mlock(), mprotect(), madvise(). File
descriptor approach allows explicit and controlled sharing of the memory
areas, it allows to seal the operations. Besides, file descriptor based
memory paves the way for VMMs to remove the secret memory range from the
userspace hipervisor process, for instance QEMU. Andy Lutomirski says:
"Getting fd-backed memory into a guest will take some possibly major
work in the kernel, but getting vma-backed memory into a guest without
mapping it in the host user address space seems much, much worse."
memfd_secret() is made a dedicated system call rather than an extension to
memfd_create() because it's purpose is to allow the user to create more
secure memory mappings rather than to simply allow file based access to
the memory. Nowadays a new system call cost is negligible while it is way
simpler for userspace to deal with a clear-cut system calls than with a
multiplexer or an overloaded syscall. Moreover, the initial
implementation of memfd_secret() is completely distinct from
memfd_create() so there is no much sense in overloading memfd_create() to
begin with. If there will be a need for code sharing between these
implementation it can be easily achieved without a need to adjust user
visible APIs.
The secret memory remains accessible in the process context using uaccess
primitives, but it is not exposed to the kernel otherwise; secret memory
areas are removed from the direct map and functions in the
follow_page()/get_user_page() family will refuse to return a page that
belongs to the secret memory area.
Once there will be a use case that will require exposing secretmem to the
kernel it will be an opt-in request in the system call flags so that user
would have to decide what data can be exposed to the kernel.
Removing of the pages from the direct map may cause its fragmentation on
architectures that use large pages to map the physical memory which
affects the system performance. However, the original Kconfig text for
CONFIG_DIRECT_GBPAGES said that gigabyte pages in the direct map "... can
improve the kernel's performance a tiny bit ..." (commit 00d1c5e05736
("x86: add gbpages switches")) and the recent report [1] showed that "...
although 1G mappings are a good default choice, there is no compelling
evidence that it must be the only choice". Hence, it is sufficient to
have secretmem disabled by default with the ability of a system
administrator to enable it at boot time.
Pages in the secretmem regions are unevictable and unmovable to avoid
accidental exposure of the sensitive data via swap or during page
migration.
Since the secretmem mappings are locked in memory they cannot exceed
RLIMIT_MEMLOCK. Since these mappings are already locked independently
from mlock(), an attempt to mlock()/munlock() secretmem range would fail
and mlockall()/munlockall() will ignore secretmem mappings.
However, unlike mlock()ed memory, secretmem currently behaves more like
long-term GUP: secretmem mappings are unmovable mappings directly consumed
by user space. With default limits, there is no excessive use of
secretmem and it poses no real problem in combination with
ZONE_MOVABLE/CMA, but in the future this should be addressed to allow
balanced use of large amounts of secretmem along with ZONE_MOVABLE/CMA.
A page that was a part of the secret memory area is cleared when it is
freed to ensure the data is not exposed to the next user of that page.
The following example demonstrates creation of a secret mapping (error
handling is omitted):
Mike Rapoport [Thu, 8 Jul 2021 01:07:59 +0000 (18:07 -0700)]
set_memory: allow querying whether set_direct_map_*() is actually enabled
On arm64, set_direct_map_*() functions may return 0 without actually
changing the linear map. This behaviour can be controlled using kernel
parameters, so we need a way to determine at runtime whether calls to
set_direct_map_invalid_noflush() and set_direct_map_default_noflush() have
any effect.
Extend set_memory API with can_set_direct_map() function that allows
checking if calling set_direct_map_*() will actually change the page
table, replace several occurrences of open coded checks in arm64 with the
new function and provide a generic stub for architectures that always
modify page tables upon calls to set_direct_map APIs.
Mike Rapoport [Thu, 8 Jul 2021 01:07:54 +0000 (18:07 -0700)]
riscv/Kconfig: make direct map manipulation options depend on MMU
ARCH_HAS_SET_DIRECT_MAP and ARCH_HAS_SET_MEMORY configuration options have
no meaning when CONFIG_MMU is disabled and there is no point to enable
them for the nommu case.
Add an explicit dependency on MMU for these options.
Mike Rapoport [Thu, 8 Jul 2021 01:07:50 +0000 (18:07 -0700)]
mmap: make mlock_future_check() global
Patch series "mm: introduce memfd_secret system call to create "secret" memory areas", v20.
This is an implementation of "secret" mappings backed by a file
descriptor.
The file descriptor backing secret memory mappings is created using a
dedicated memfd_secret system call The desired protection mode for the
memory is configured using flags parameter of the system call. The mmap()
of the file descriptor created with memfd_secret() will create a "secret"
memory mapping. The pages in that mapping will be marked as not present
in the direct map and will be present only in the page table of the owning
mm.
Although normally Linux userspace mappings are protected from other users,
such secret mappings are useful for environments where a hostile tenant is
trying to trick the kernel into giving them access to other tenants
mappings.
It's designed to provide the following protections:
* Enhanced protection (in conjunction with all the other in-kernel
attack prevention systems) against ROP attacks. Seceretmem makes
"simple" ROP insufficient to perform exfiltration, which increases the
required complexity of the attack. Along with other protections like
the kernel stack size limit and address space layout randomization which
make finding gadgets is really hard, absence of any in-kernel primitive
for accessing secret memory means the one gadget ROP attack can't work.
Since the only way to access secret memory is to reconstruct the missing
mapping entry, the attacker has to recover the physical page and insert
a PTE pointing to it in the kernel and then retrieve the contents. That
takes at least three gadgets which is a level of difficulty beyond most
standard attacks.
* Prevent cross-process secret userspace memory exposures. Once the
secret memory is allocated, the user can't accidentally pass it into the
kernel to be transmitted somewhere. The secreremem pages cannot be
accessed via the direct map and they are disallowed in GUP.
* Harden against exploited kernel flaws. In order to access secretmem,
a kernel-side attack would need to either walk the page tables and
create new ones, or spawn a new privileged uiserspace process to perform
secrets exfiltration using ptrace.
In the future the secret mappings may be used as a mean to protect guest
memory in a virtual machine host.
For demonstration of secret memory usage we've created a userspace library
that does two things: the first is act as a preloader for openssl to
redirect all the OPENSSL_malloc calls to secret memory meaning any secret
keys get automatically protected this way and the other thing it does is
expose the API to the user who needs it. We anticipate that a lot of the
use cases would be like the openssl one: many toolkits that deal with
secret keys already have special handling for the memory to try to give
them greater protection, so this would simply be pluggable into the
toolkits without any need for user application modification.
Hiding secret memory mappings behind an anonymous file allows usage of the
page cache for tracking pages allocated for the "secret" mappings as well
as using address_space_operations for e.g. page migration callbacks.
The anonymous file may be also used implicitly, like hugetlb files, to
implement mmap(MAP_SECRET) and use the secret memory areas with "native"
mm ABIs in the future.
Removing of the pages from the direct map may cause its fragmentation on
architectures that use large pages to map the physical memory which
affects the system performance. However, the original Kconfig text for
CONFIG_DIRECT_GBPAGES said that gigabyte pages in the direct map "... can
improve the kernel's performance a tiny bit ..." (commit 00d1c5e05736
("x86: add gbpages switches")) and the recent report [1] showed that "...
although 1G mappings are a good default choice, there is no compelling
evidence that it must be the only choice". Hence, it is sufficient to
have secretmem disabled by default with the ability of a system
administrator to enable it at boot time.
In addition, there is also a long term goal to improve management of the
direct map.
Oliver Glitta [Thu, 8 Jul 2021 01:07:47 +0000 (18:07 -0700)]
mm/slub: use stackdepot to save stack trace in objects
Many stack traces are similar so there are many similar arrays.
Stackdepot saves each unique stack only once.
Replace field addrs in struct track with depot_stack_handle_t handle. Use
stackdepot to save stack trace.
The benefits are smaller memory overhead and possibility to aggregate
per-cache statistics in the future using the stackdepot handle instead of
matching stacks manually.
ld.lld warns that the '.modinfo' section is not currently handled:
ld.lld: warning: kernel/built-in.a(workqueue.o):(.modinfo) is being placed in '.modinfo'
ld.lld: warning: kernel/built-in.a(printk/printk.o):(.modinfo) is being placed in '.modinfo'
ld.lld: warning: kernel/built-in.a(irq/spurious.o):(.modinfo) is being placed in '.modinfo'
ld.lld: warning: kernel/built-in.a(rcu/update.o):(.modinfo) is being placed in '.modinfo'
The '.modinfo' section was added in commit 898490c010b5 ("moduleparam:
Save information about built-in modules in separate file") to the DISCARDS
macro but Hexagon has never used that macro. The unification of DISCARDS
happened in commit 023bf6f1b8bf ("linker script: unify usage of discard
definition") in 2009, prior to Hexagon being added in 2011.
Switch Hexagon over to the DISCARDS macro so that anything that is
expected to be discarded gets discarded.
hexagon: handle {,SOFT}IRQENTRY_TEXT in linker script
Patch series "hexagon: Fix build error with CONFIG_STACKDEPOT and select CONFIG_ARCH_WANT_LD_ORPHAN_WARN".
This series fixes an error with ARCH=hexagon that was pointed out by the
patch "mm/slub: use stackdepot to save stack trace in objects".
The first patch fixes that error by handling the '.irqentry.text' and
'.softirqentry.text' sections.
The second patch switches Hexagon over to the common DISCARDS macro, which
should have been done when Hexagon was merged into the tree to match
commit 023bf6f1b8bf ("linker script: unify usage of discard definition").
The third patch selects CONFIG_ARCH_WANT_LD_ORPHAN_WARN so that something
like this does not happen again.
This patch (of 3):
Patch "mm/slub: use stackdepot to save stack trace in objects" in -mm
selects CONFIG_STACKDEPOT when CONFIG_STACKTRACE_SUPPORT is selected and
CONFIG_STACKDEPOT requires IRQENTRY_TEXT and SOFTIRQENTRY_TEXT to be
handled after commit 505a0ef15f96 ("kasan: stackdepot: move
filter_irq_stacks() to stackdepot.c") due to the use of the
__{,soft}irqentry_text_{start,end} section symbols. If those sections are
not handled, the build is broken.
$ make ARCH=hexagon CROSS_COMPILE=hexagon-linux- LLVM=1 LLVM_IAS=1 defconfig all
...
ld.lld: error: undefined symbol: __irqentry_text_start
>>> referenced by stackdepot.c
>>> stackdepot.o:(filter_irq_stacks) in archive lib/built-in.a
>>> referenced by stackdepot.c
>>> stackdepot.o:(filter_irq_stacks) in archive lib/built-in.a
ld.lld: error: undefined symbol: __irqentry_text_end
>>> referenced by stackdepot.c
>>> stackdepot.o:(filter_irq_stacks) in archive lib/built-in.a
>>> referenced by stackdepot.c
>>> stackdepot.o:(filter_irq_stacks) in archive lib/built-in.a
ld.lld: error: undefined symbol: __softirqentry_text_start
>>> referenced by stackdepot.c
>>> stackdepot.o:(filter_irq_stacks) in archive lib/built-in.a
>>> referenced by stackdepot.c
>>> stackdepot.o:(filter_irq_stacks) in archive lib/built-in.a
ld.lld: error: undefined symbol: __softirqentry_text_end
>>> referenced by stackdepot.c
>>> stackdepot.o:(filter_irq_stacks) in archive lib/built-in.a
>>> referenced by stackdepot.c
>>> stackdepot.o:(filter_irq_stacks) in archive lib/built-in.a
...
Add these sections to the Hexagon linker script so the build continues to
work. ld.lld's orphan section warning would have caught this prior to the
-mm commit mentioned above:
ld.lld: warning: kernel/built-in.a(softirq.o):(.softirqentry.text) is being placed in '.softirqentry.text'
ld.lld: warning: kernel/built-in.a(softirq.o):(.softirqentry.text) is being placed in '.softirqentry.text'
ld.lld: warning: kernel/built-in.a(softirq.o):(.softirqentry.text) is being placed in '.softirqentry.text'
Zhen Lei [Thu, 8 Jul 2021 01:07:34 +0000 (18:07 -0700)]
lib: fix spelling mistakes in header files
Fix some spelling mistakes in comments found by "codespell":
Hoever ==> However
poiter ==> pointer
representaion ==> representation
uppon ==> upon
independend ==> independent
aquired ==> acquired
mis-match ==> mismatch
scrach ==> scratch
struture ==> structure
Analagous ==> Analogous
interation ==> iteration
And some were discovered manually by Joe Perches and Christoph Lameter:
stroed ==> stored
arch independent ==> an architecture independent
A example structure for ==> Example structure for
We must properly handle an errors when we increase the rlimit counter
and the ucounts reference counter. We have to this with RCU protection
to prevent possible use-after-free that could occur due to concurrent
put_cred_rcu().
The following reproducer triggers the problem:
$ cat testcase.sh
case "${STEP:-0}" in
0)
ulimit -Si 1
ulimit -Hi 1
STEP=1 unshare -rU "$0"
killall sleep
;;
1)
for i in 1 2 3 4 5; do unshare -rU sleep 5 & done
;;
esac
with the KASAN report being along the lines of
BUG: KASAN: use-after-free in put_ucounts+0x17/0xa0
Write of size 4 at addr ffff8880045f031c by task swapper/2/0
Memory state around the buggy address: ffff8880045f0200: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb ffff8880045f0280: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
>ffff8880045f0300: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
^ ffff8880045f0380: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc ffff8880045f0400: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
==================================================================
Disabling lock debugging due to kernel taint
NFSv4/pNFS: Don't call _nfs4_pnfs_v3_ds_connect multiple times
After we grab the lock in nfs4_pnfs_ds_connect(), there is no check for
whether or not ds->ds_clp has already been initialised, so we can end up
adding the same transports multiple times.
Fixes: fc821d59209d ("pnfs/NFSv4.1: Add multipath capabilities to pNFS flexfiles servers over NFSv3") Signed-off-by: Trond Myklebust <[email protected]>
NFSv4/pnfs: Fix layoutget behaviour after invalidation
If the layout gets invalidated, we should wait for any outstanding
layoutget requests for that layout to complete, and we should resend
them only after re-establishing the layout stateid.
If we have multiple outstanding layoutget requests, the current code to
update the layout barrier assumes that the outstanding layout stateids
are updated in order. That's not necessarily the case.
Instead of using the value of lo->plh_outstanding as a guesstimate for
the window of values we need to accept, just wait to update the window
until we're processing the last one. The intention here is just to
ensure that we don't process 2^31 seqid updates without also updating
the barrier.
Fixes: 1bcf34fdac5f ("pNFS/NFSv4: Update the layout barrier when we schedule a layoutreturn") Signed-off-by: Trond Myklebust <[email protected]>
Dave Wysochanski [Tue, 29 Jun 2021 17:13:57 +0000 (13:13 -0400)]
NFS: Fix fscache read from NFS after cache error
Earlier commits refactored some NFS read code and removed
nfs_readpage_async(), but neglected to properly fixup
nfs_readpage_from_fscache_complete(). The code path is
only hit when something unusual occurs with the cachefiles
backing filesystem, such as an IO error or while a cookie
is being invalidated.
Mark page with PG_checked if fscache IO completes in error,
unlock the page, and let the VM decide to re-issue based on
PG_uptodate. When the VM reissues the readpage, PG_checked
allows us to skip over fscache and read from the server.
Dave Wysochanski [Tue, 29 Jun 2021 09:11:28 +0000 (05:11 -0400)]
NFS: Ensure nfs_readpage returns promptly when internal error occurs
A previous refactoring of nfs_readpage() might end up calling
wait_on_page_locked_killable() even if readpage_async_filler() failed
with an internal error and pg_error was non-zero (for example, if
nfs_create_request() failed). In the case of an internal error,
skip over wait_on_page_locked_killable() as this is only needed
when the read is sent and an error occurs during completion handling.
Once a transport has been put offline, this transport can be also
removed from the list of transports. Any tasks that have been stuck
on this transport would find the next available active transport
and be re-tried. This transport would be removed from the xprt_switch
list and freed.
NFSv4.1 identify and mark RPC tasks that can move between transports
In preparation for when we can re-try a task on a different transport,
identify and mark such RPC tasks as moveable. Only 4.1+ operarations can
be re-tried on a different transport.
sunrpc: provide transport info in the sysfs directory
Allow to query transport's attributes. Currently showing following
fields of the rpc_xprt structure: state, last_used, cong, cwnd,
max_reqs, min_reqs, num_reqs, sizes of queues binding, sending,
pending, backlog.
Using sysfs's xprt_state attribute, mark a particular transport offline.
It will not be picked during the round-robin selection. It's not allowed
to take the main (1st created transport associated with the rpc_client)
offline. Also bring a transport back online via sysfs by writing "online"
and that would allow for this transport to be picked during the round-
robin selection.
sunrpc: add add sysfs directory per xprt under each xprt_switch
Add individual transport directories under each transport switch
group. For instance, for each nconnect=X connections there will be
a transport directory. Naming conventions also identifies transport
type -- xprt-<id>-<type> where type is udp, tcp, rdma, local, bc.
sunrpc: add a symlink from rpc-client directory to the xprt_switch
An rpc client uses a transport switch and one ore more transports
associated with that switch. Since transports are shared among
rpc clients, create a symlink into the xprt_switch directory
instead of duplicating entries under each rpc client.
powerpc/preempt: Don't touch the idle task's preempt_count during hotplug
Powerpc currently resets a CPU's idle task preempt_count to 0 before
said task starts executing the secondary startup routine (and becomes an
idle task proper).
This conflicts with commit f1a0a376ca0c ("sched/core: Initialize the
idle task with preemption disabled").
which initializes all of the idle tasks' preempt_count to
PREEMPT_DISABLED during smp_init(). Note that this was superfluous
before said commit, as back then the hotplug machinery would invoke
init_idle() via idle_thread_get(), which would have already reset the
CPU's idle task's preempt_count to PREEMPT_ENABLED.
Sven Schnelle [Fri, 25 Jun 2021 12:50:08 +0000 (14:50 +0200)]
s390/vdso: add minimal compat vdso
Add a small vdso for 31 bit compat application that provides
trampolines for calls to sigreturn,rt_sigreturn,syscall_restart.
This is requird for moving these syscalls away from the signal
frame to the vdso. Note that this patch effectively disables
CONFIG_COMPAT when using clang to compile the kernel. clang
doesn't support 31 bit mode.
We want to redirect sigreturn and restart_syscall to the vdso. However,
the kernel cannot parse the ELF vdso file, so we need to generate header
files which contain the offsets of the syscall instructions in the vdso
page.
Sven Schnelle [Fri, 25 Jun 2021 12:48:28 +0000 (14:48 +0200)]
s390/vdso64: add sigreturn,rt_sigreturn and restart_syscall
Add minimalistic trampolines to vdso64 so we can return from signal
without using the stack which requires pgm check handler hacks when
NX is enabled.
restart_syscall will be called from vdso to work around the architectural
limitation that the syscall number might be encoded in the svc instruction,
and therefore can not be changed.
Using register asm statements has been proven to be very error prone,
especially when using code instrumentation where gcc may add function
calls, which clobbers register contents in an unexpected way.
Therefore get rid of register asm statements in ap code. There are also
potential bugs, depending on inline decisions of the compiler.
E.g. for:
static inline struct ap_queue_status ap_tapq(ap_qid_t qid, unsigned long *info)
{
register unsigned long reg0 asm ("0") = qid;
register struct ap_queue_status reg1 asm ("1");
register unsigned long reg2 asm ("2");
In case of KCOV the "if (info)" line could cause a generated function
call, which could clobber the contents of both reg2, and reg1.
Similar can happen in case of KASAN for the "*info = reg2" line.
Even though compilers will likely inline the function and optimize
things away, this is not guaranteed.
To get rid of this bug class, simply get rid of register asm constructs.
Note: The inline function ap_dqap() will be handled in a
separate patch because this one requires an addressing of the
odd register of a register pair (which is done with %N[xxx] in
the assembler code) and that's currently not supported by clang.
Sven Schnelle [Tue, 6 Jul 2021 18:24:58 +0000 (20:24 +0200)]
s390/irq: remove HAVE_IRQ_EXIT_ON_IRQ_STACK
This is no longer true since we switched to generic entry. The code
switches to the IRQ stack before calling do_IRQ, but switches back
to the kernel stack before calling irq_exit().
s390/traps: do not test MONITOR CALL without CONFIG_BUG
tinyconfig fails to boot, because without CONFIG_BUG report_bug()
always returns BUG_TRAP_TYPE_BUG, which causes mc 0,0 in
test_monitor_call() to panic. Fix by skipping the test without
CONFIG_BUG.
s390/ap: Rework ap_dqap to deal with messages greater than recv buffer
Rework of the ap_dqap() inline function with the dqap inline assembler
invocation and the caller code in ap_queue.c to be able to handle
replies which exceed the receive buffer size.
ap_dqap() now provides two additional parameters to handle together
with the caller the case where a reply in the firmware queue entry
exceeds the given message buffer size. It depends on the caller how to
exactly handle this. The behavior implemented now by ap_sm_recv() in
ap_queue.c is to simple purge this entry from the firmware queue and
let the caller 'receive' a -EMSGSIZE for the request without
delivering any reply data - not even a truncated reply message.
However, the reworked ap_dqap() could now get invoked in a way that
the message is received in multiple parts and the caller assembles the
parts into one reply message.
The function jbd2_journal_unregister_shrinker() was getting called
twice when the file system was getting unmounted. On Power and ARM
platforms this was causing kernel crash when unmounting the file
system, when a percpu_counter was destroyed twice.
Fix this by removing jbd2_journal_[un]register_shrinker() functions,
and inlining the shrinker setup and teardown into
journal_init_common() and jbd2_journal_destroy(). This means that
ext4 and ocfs2 now no longer need to know about registering and
unregistering jbd2's shrinker.
Also, while we're at it, rename the percpu counter from
j_jh_shrink_count to j_checkpoint_jh_count, since this makes it
clearer what this counter is intended to track.
ext4: fix possible UAF when remounting r/o a mmp-protected file system
After commit 618f003199c6 ("ext4: fix memory leak in
ext4_fill_super"), after the file system is remounted read-only, there
is a race where the kmmpd thread can exit, causing sbi->s_mmp_tsk to
point at freed memory, which the call to ext4_stop_mmpd() can trip
over.
Fix this by only allowing kmmpd() to exit when it is stopped via
ext4_stop_mmpd().
virtio-mem: simplify high-level unplug handling in Big Block Mode
Let's simplify high-level big block selection when unplugging in
Big Block Mode.
Combine handling of offline and online blocks. We can get rid of
virtio_mem_bbm_bb_is_offline() and simply use
virtio_mem_bbm_offline_remove_and_unplug_bb(), as that already tolerates
offline parts.
We can race with concurrent onlining/offlining either way, so we don;t
have to be super correct by failing if an offline big block we'd like to
unplug just got (partially) onlined.
virtio-mem: prioritize unplug from ZONE_MOVABLE in Sub Block Mode
Until now, memory provided by a single virtio-mem device was usually
either onlined completely to ZONE_MOVABLE (online_movable) or to
ZONE_NORMAL (online_kernel); however, that will change in the future.
There are two reasons why we want to track to which zone a memory blocks
belongs to and prioritize ZONE_MOVABLE blocks:
1) Memory managed by ZONE_MOVABLE can more likely get unplugged, therefore,
resulting in a faster memory hotunplug process. Further, we can more
reliably unplug and remove complete memory blocks, removing metadata
allocated for the whole memory block.
2) We want to avoid corner cases where unplugging with the current scheme
(highest to lowest address) could result in accidential zone imbalances,
whereby we remove too much ZONE_NORMAL memory for ZONE_MOVABLE memory
of the same device.
Let's track the zone via memory block states and try unplug from
ZONE_MOVABLE first. Rename VIRTIO_MEM_SBM_MB_ONLINE* to
VIRTIO_MEM_SBM_MB_KERNEL* to avoid even longer state names.
In commit 27f852795a06 ("virtio-mem: don't special-case ZONE_MOVABLE"),
we removed slightly similar tracking for fully plugged memory blocks to
support unplugging from ZONE_MOVABLE at all -- as we didn't allow partially
plugged memory blocks in ZONE_MOVABLE before that. That commit already
mentioned "In the future, we might want to remember the zone again and use
the information when (un)plugging memory."
virtio-mem: simplify high-level unplug handling in Sub Block Mode
Let's simplify by introducing a new virtio_mem_sbm_unplug_any_sb(),
similar to virtio_mem_sbm_plug_any_sb(), to simplify high-level memory
block selection when unplugging in Sub Block Mode.
Rename existing virtio_mem_sbm_unplug_any_sb() to
virtio_mem_sbm_unplug_any_sb_raw().
The only change is that we now temporarily unlock the hotplug mutex around
cond_resched() when processing offline memory blocks, which doesn't
make a real difference as we already have to temporarily unlock in
virtio_mem_sbm_unplug_any_sb_offline() when removing a memory block.
virtio-mem: simplify high-level plug handling in Sub Block Mode
Let's simplify high-level memory block selection when plugging in Sub
Block Mode.
No need for two separate loops when selecting memory blocks for plugging
memory. Avoid passing the "online" state by simply obtaining the state
in virtio_mem_sbm_plug_any_sb().
virtio-mem: don't read big block size in Sub Block Mode
We are reading a Big Block Mode value while in Sub Block Mode
when initializing. Fortunately, vm->bbm.bb_size maps to some counter
in the vm->sbm.mb_count array, which is 0 at that point in time.
No harm done; still, this was unintended and is not future-proof.
Eli Cohen [Wed, 2 Jun 2021 02:15:36 +0000 (10:15 +0800)]
virtio/vdpa: clear the virtqueue state during probe
Clear the available index as part of the initialization process to
clear and values that might be left from previous usage of the device.
For example, if the device was previously used by vhost_vdpa and now
probed by vhost_vdpa, you want to start with indices.
Jason Wang [Wed, 2 Jun 2021 02:15:35 +0000 (10:15 +0800)]
vp_vdpa: allow set vq state to initial state after reset
We used to fail the set_vq_state() since it was not supported yet by
the virtio spec. But if the bus tries to set the state which is equal
to the device initial state after reset, we can let it go.
This is a must for virtio_vdpa() to set vq state during probe which is
required for some vDPA parents.
Jason Wang [Wed, 2 Jun 2021 02:15:33 +0000 (10:15 +0800)]
vdpa: support packed virtqueue for set/get_vq_state()
This patch extends the vdpa_vq_state to support packed virtqueue
state which is basically the device/driver ring wrap counters and the
avail and used index. This will be used for the virito-vdpa support
for the packed virtqueue and the future vhost/vhost-vdpa support for
the packed virtqueue.
Jason Wang [Fri, 4 Jun 2021 05:53:50 +0000 (13:53 +0800)]
virtio-ring: store DMA metadata in desc_extra for split virtqueue
For split virtqueue, we used to depend on the address, length and
flags stored in the descriptor ring for DMA unmapping. This is unsafe
for the case since the device can manipulate the behavior of virtio
driver, IOMMU drivers and swiotlb.
For safety, maintain the DMA address, DMA length, descriptor flags and
next filed of the non indirect descriptors in vring_desc_state_extra
when DMA API is used for virtio as we did for packed virtqueue and use
those metadata for performing DMA operations. Indirect descriptors
should be safe since they are using streaming mappings.
With this the descriptor ring is write only form the view of the
driver.
This slight increase the footprint of the drive but it's not noticed
through pktgen (64B) test and netperf test in the case of virtio-net.
projects / linux.git / log