This series focuses on minor clean-ups and performance optimizations across
sysfs, documentation, debugfs, tracepoints, slab allocation, and GC.
Furthermore, it resolves several corner-case bugs caught by xfstests, as
well as issues related to 16KB page support and f2fs_enable_checkpoint.
Enhancement:
- wrap ASCII tables in literal blocks to fix LaTeX build
- optimize trace_f2fs_write_checkpoint with enums
- support to show curseg.next_blkoff in debugfs
- add a sysfs entry to show max open zones
- add fadvise tracepoint
- use global inline_xattr_slab instead of per-sb slab cache
- set default valid_thresh_ratio to 80 for zoned devices
- maintain one time GC mode is enabled during whole zoned GC cycle
Bug fix:
- ensure node page reads complete before f2fs_put_super() finishes
- fix to not account invalid blocks in get_left_section_blocks()
- revert summary entry count from 2048 to 512 in 16kb block support
- fix to detect recoverable inode during dryrun of find_fsync_dnodes()
- fix age extent cache insertion skip on counter overflow
- Add sanity checks before unlinking and loading inodes
- ensure minimum trim granularity accounts for all devices
- block cache/dio write during f2fs_enable_checkpoint()
- fix to propagate error from f2fs_enable_checkpoint()
- invalidate dentry cache on failed whiteout creation
- fix to avoid updating compression context during writeback
- fix to avoid updating zero-sized extent in extent cache
- fix to avoid potential deadlock
-----BEGIN PGP SIGNATURE-----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=4ZOp
-----END PGP SIGNATURE-----
Merge tag 'f2fs-for-6.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs
Pull f2fs updates from Jaegeuk Kim:
"This series focuses on minor clean-ups and performance optimizations
across sysfs, documentation, debugfs, tracepoints, slab allocation,
and GC. Furthermore, it resolves several corner-case bugs caught by
xfstests, as well as issues related to 16KB page support and
f2fs_enable_checkpoint.
Enhancement:
- wrap ASCII tables in literal blocks to fix LaTeX build
- optimize trace_f2fs_write_checkpoint with enums
- support to show curseg.next_blkoff in debugfs
- add a sysfs entry to show max open zones
- add fadvise tracepoint
- use global inline_xattr_slab instead of per-sb slab cache
- set default valid_thresh_ratio to 80 for zoned devices
- maintain one time GC mode is enabled during whole zoned GC cycle
Bug fix:
- ensure node page reads complete before f2fs_put_super() finishes
- do not account invalid blocks in get_left_section_blocks()
- revert summary entry count from 2048 to 512 in 16kb block support
- detect recoverable inode during dryrun of find_fsync_dnodes()
- fix age extent cache insertion skip on counter overflow
- add sanity checks before unlinking and loading inodes
- ensure minimum trim granularity accounts for all devices
- block cache/dio write during f2fs_enable_checkpoint()
- propagate error from f2fs_enable_checkpoint()
- invalidate dentry cache on failed whiteout creation
- avoid updating compression context during writeback
- avoid updating zero-sized extent in extent cache
- avoid potential deadlock"
* tag 'f2fs-for-6.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs: (39 commits)
f2fs: ignore discard return value
f2fs: optimize trace_f2fs_write_checkpoint with enums
f2fs: fix to not account invalid blocks in get_left_section_blocks()
f2fs: support to show curseg.next_blkoff in debugfs
docs: f2fs: wrap ASCII tables in literal blocks to fix LaTeX build
f2fs: expand scalability of f2fs mount option
f2fs: change default schedule timeout value
f2fs: introduce f2fs_schedule_timeout()
f2fs: use memalloc_retry_wait() as much as possible
f2fs: add a sysfs entry to show max open zones
f2fs: wrap all unusable_blocks_per_sec code in CONFIG_BLK_DEV_ZONED
f2fs: simplify list initialization in f2fs_recover_fsync_data()
f2fs: revert summary entry count from 2048 to 512 in 16kb block support
f2fs: fix to detect recoverable inode during dryrun of find_fsync_dnodes()
f2fs: fix return value of f2fs_recover_fsync_data()
f2fs: add fadvise tracepoint
f2fs: fix age extent cache insertion skip on counter overflow
f2fs: Add sanity checks before unlinking and loading inodes
f2fs: Rename f2fs_unlink exit label
f2fs: ensure minimum trim granularity accounts for all devices
...
- Support for userspace handling of synchronous external aborts (SEAs),
allowing the VMM to potentially handle the abort in a non-fatal
manner.
- Large rework of the VGIC's list register handling with the goal of
supporting more active/pending IRQs than available list registers in
hardware. In addition, the VGIC now supports EOImode==1 style
deactivations for IRQs which may occur on a separate vCPU than the
one that acked the IRQ.
- Support for FEAT_XNX (user / privileged execute permissions) and
FEAT_HAF (hardware update to the Access Flag) in the software page
table walkers and shadow MMU.
- Allow page table destruction to reschedule, fixing long need_resched
latencies observed when destroying a large VM.
- Minor fixes to KVM and selftests
Loongarch:
- Get VM PMU capability from HW GCFG register.
- Add AVEC basic support.
- Use 64-bit register definition for EIOINTC.
- Add KVM timer test cases for tools/selftests.
RISC/V:
- SBI message passing (MPXY) support for KVM guest
- Give a new, more specific error subcode for the case when in-kernel
AIA virtualization fails to allocate IMSIC VS-file
- Support KVM_DIRTY_LOG_INITIALLY_SET, enabling dirty log gradually
in small chunks
- Fix guest page fault within HLV* instructions
- Flush VS-stage TLB after VCPU migration for Andes cores
s390:
- Always allocate ESCA (Extended System Control Area), instead of
starting with the basic SCA and converting to ESCA with the
addition of the 65th vCPU. The price is increased number of
exits (and worse performance) on z10 and earlier processor;
ESCA was introduced by z114/z196 in 2010.
- VIRT_XFER_TO_GUEST_WORK support
- Operation exception forwarding support
- Cleanups
x86:
- Skip the costly "zap all SPTEs" on an MMIO generation wrap if MMIO SPTE
caching is disabled, as there can't be any relevant SPTEs to zap.
- Relocate a misplaced export.
- Fix an async #PF bug where KVM would clear the completion queue when the
guest transitioned in and out of paging mode, e.g. when handling an SMI and
then returning to paged mode via RSM.
- Leave KVM's user-return notifier registered even when disabling
virtualization, as long as kvm.ko is loaded. On reboot/shutdown, keeping
the notifier registered is ok; the kernel does not use the MSRs and the
callback will run cleanly and restore host MSRs if the CPU manages to
return to userspace before the system goes down.
- Use the checked version of {get,put}_user().
- Fix a long-lurking bug where KVM's lack of catch-up logic for periodic APIC
timers can result in a hard lockup in the host.
- Revert the periodic kvmclock sync logic now that KVM doesn't use a
clocksource that's subject to NTP corrections.
- Clean up KVM's handling of MMIO Stale Data and L1TF, and bury the latter
behind CONFIG_CPU_MITIGATIONS.
- Context switch XCR0, XSS, and PKRU outside of the entry/exit fast path;
the only reason they were handled in the fast path was to paper of a bug
in the core #MC code, and that has long since been fixed.
- Add emulator support for AVX MOV instructions, to play nice with emulated
devices whose guest drivers like to access PCI BARs with large multi-byte
instructions.
x86 (AMD):
- Fix a few missing "VMCB dirty" bugs.
- Fix the worst of KVM's lack of EFER.LMSLE emulation.
- Add AVIC support for addressing 4k vCPUs in x2AVIC mode.
- Fix incorrect handling of selective CR0 writes when checking intercepts
during emulation of L2 instructions.
- Fix a currently-benign bug where KVM would clobber SPEC_CTRL[63:32] on
VMRUN and #VMEXIT.
- Fix a bug where KVM corrupt the guest code stream when re-injecting a soft
interrupt if the guest patched the underlying code after the VM-Exit, e.g.
when Linux patches code with a temporary INT3.
- Add KVM_X86_SNP_POLICY_BITS to advertise supported SNP policy bits to
userspace, and extend KVM "support" to all policy bits that don't require
any actual support from KVM.
x86 (Intel):
- Use the root role from kvm_mmu_page to construct EPTPs instead of the
current vCPU state, partly as worthwhile cleanup, but mostly to pave the
way for tracking per-root TLB flushes, and elide EPT flushes on pCPU
migration if the root is clean from a previous flush.
- Add a few missing nested consistency checks.
- Rip out support for doing "early" consistency checks via hardware as the
functionality hasn't been used in years and is no longer useful in general;
replace it with an off-by-default module param to WARN if hardware fails
a check that KVM does not perform.
- Fix a currently-benign bug where KVM would drop the guest's SPEC_CTRL[63:32]
on VM-Enter.
- Misc cleanups.
- Overhaul the TDX code to address systemic races where KVM (acting on behalf
of userspace) could inadvertantly trigger lock contention in the TDX-Module;
KVM was either working around these in weird, ugly ways, or was simply
oblivious to them (though even Yan's devilish selftests could only break
individual VMs, not the host kernel)
- Fix a bug where KVM could corrupt a vCPU's cpu_list when freeing a TDX vCPU,
if creating said vCPU failed partway through.
- Fix a few sparse warnings (bad annotation, 0 != NULL).
- Use struct_size() to simplify copying TDX capabilities to userspace.
- Fix a bug where TDX would effectively corrupt user-return MSR values if the
TDX Module rejects VP.ENTER and thus doesn't clobber host MSRs as expected.
Selftests:
- Fix a math goof in mmu_stress_test when running on a single-CPU system/VM.
- Forcefully override ARCH from x86_64 to x86 to play nice with specifying
ARCH=x86_64 on the command line.
- Extend a bunch of nested VMX to validate nested SVM as well.
- Add support for LA57 in the core VM_MODE_xxx macro, and add a test to
verify KVM can save/restore nested VMX state when L1 is using 5-level
paging, but L2 is not.
- Clean up the guest paging code in anticipation of sharing the core logic for
nested EPT and nested NPT.
guest_memfd:
- Add NUMA mempolicy support for guest_memfd, and clean up a variety of
rough edges in guest_memfd along the way.
- Define a CLASS to automatically handle get+put when grabbing a guest_memfd
from a memslot to make it harder to leak references.
- Enhance KVM selftests to make it easer to develop and debug selftests like
those added for guest_memfd NUMA support, e.g. where test and/or KVM bugs
often result in hard-to-debug SIGBUS errors.
- Misc cleanups.
Generic:
- Use the recently-added WQ_PERCPU when creating the per-CPU workqueue for
irqfd cleanup.
- Fix a goof in the dirty ring documentation.
- Fix choice of target for directed yield across different calls to
kvm_vcpu_on_spin(); the function was always starting from the first
vCPU instead of continuing the round-robin search.
-----BEGIN PGP SIGNATURE-----
iQFIBAABCgAyFiEE8TM4V0tmI4mGbHaCv/vSX3jHroMFAmkvMa8UHHBib256aW5p
QHJlZGhhdC5jb20ACgkQv/vSX3jHroMlFwf+Ow7zOYUuELSQ+Jn+hOYXiCNrdBDx
ZamvMU8kLPr7XX0Zog6HgcMm//qyA6k5nSfqCjfsQZrIhRA/gWJ61jz1OX/Jxq18
pJ9Vz6epnEPYiOtBwz+v8OS8MqDqVNzj2i6W1/cLPQE50c1Hhw64HWS5CSxDQiHW
A7PVfl5YU12lW1vG3uE0sNESDt4Eh/spNM17iddXdF4ZUOGublserjDGjbc17E7H
8BX3DkC2plqkJKwtjg0ae62hREkITZZc7RqsnftUkEhn0N0H9+rb6NKUyzIVh9NZ
bCtCjtrKN9zfZ0Mujnms3ugBOVqNIputu/DtPnnFKXtXWSrHrgGSNv5ewA==
=PEcw
-----END PGP SIGNATURE-----
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Pull KVM updates from Paolo Bonzini:
"ARM:
- Support for userspace handling of synchronous external aborts
(SEAs), allowing the VMM to potentially handle the abort in a
non-fatal manner
- Large rework of the VGIC's list register handling with the goal of
supporting more active/pending IRQs than available list registers
in hardware. In addition, the VGIC now supports EOImode==1 style
deactivations for IRQs which may occur on a separate vCPU than the
one that acked the IRQ
- Support for FEAT_XNX (user / privileged execute permissions) and
FEAT_HAF (hardware update to the Access Flag) in the software page
table walkers and shadow MMU
- Allow page table destruction to reschedule, fixing long
need_resched latencies observed when destroying a large VM
- Minor fixes to KVM and selftests
Loongarch:
- Get VM PMU capability from HW GCFG register
- Add AVEC basic support
- Use 64-bit register definition for EIOINTC
- Add KVM timer test cases for tools/selftests
RISC/V:
- SBI message passing (MPXY) support for KVM guest
- Give a new, more specific error subcode for the case when in-kernel
AIA virtualization fails to allocate IMSIC VS-file
- Support KVM_DIRTY_LOG_INITIALLY_SET, enabling dirty log gradually
in small chunks
- Fix guest page fault within HLV* instructions
- Flush VS-stage TLB after VCPU migration for Andes cores
s390:
- Always allocate ESCA (Extended System Control Area), instead of
starting with the basic SCA and converting to ESCA with the
addition of the 65th vCPU. The price is increased number of exits
(and worse performance) on z10 and earlier processor; ESCA was
introduced by z114/z196 in 2010
- VIRT_XFER_TO_GUEST_WORK support
- Operation exception forwarding support
- Cleanups
x86:
- Skip the costly "zap all SPTEs" on an MMIO generation wrap if MMIO
SPTE caching is disabled, as there can't be any relevant SPTEs to
zap
- Relocate a misplaced export
- Fix an async #PF bug where KVM would clear the completion queue
when the guest transitioned in and out of paging mode, e.g. when
handling an SMI and then returning to paged mode via RSM
- Leave KVM's user-return notifier registered even when disabling
virtualization, as long as kvm.ko is loaded. On reboot/shutdown,
keeping the notifier registered is ok; the kernel does not use the
MSRs and the callback will run cleanly and restore host MSRs if the
CPU manages to return to userspace before the system goes down
- Use the checked version of {get,put}_user()
- Fix a long-lurking bug where KVM's lack of catch-up logic for
periodic APIC timers can result in a hard lockup in the host
- Revert the periodic kvmclock sync logic now that KVM doesn't use a
clocksource that's subject to NTP corrections
- Clean up KVM's handling of MMIO Stale Data and L1TF, and bury the
latter behind CONFIG_CPU_MITIGATIONS
- Context switch XCR0, XSS, and PKRU outside of the entry/exit fast
path; the only reason they were handled in the fast path was to
paper of a bug in the core #MC code, and that has long since been
fixed
- Add emulator support for AVX MOV instructions, to play nice with
emulated devices whose guest drivers like to access PCI BARs with
large multi-byte instructions
x86 (AMD):
- Fix a few missing "VMCB dirty" bugs
- Fix the worst of KVM's lack of EFER.LMSLE emulation
- Add AVIC support for addressing 4k vCPUs in x2AVIC mode
- Fix incorrect handling of selective CR0 writes when checking
intercepts during emulation of L2 instructions
- Fix a currently-benign bug where KVM would clobber SPEC_CTRL[63:32]
on VMRUN and #VMEXIT
- Fix a bug where KVM corrupt the guest code stream when re-injecting
a soft interrupt if the guest patched the underlying code after the
VM-Exit, e.g. when Linux patches code with a temporary INT3
- Add KVM_X86_SNP_POLICY_BITS to advertise supported SNP policy bits
to userspace, and extend KVM "support" to all policy bits that
don't require any actual support from KVM
x86 (Intel):
- Use the root role from kvm_mmu_page to construct EPTPs instead of
the current vCPU state, partly as worthwhile cleanup, but mostly to
pave the way for tracking per-root TLB flushes, and elide EPT
flushes on pCPU migration if the root is clean from a previous
flush
- Add a few missing nested consistency checks
- Rip out support for doing "early" consistency checks via hardware
as the functionality hasn't been used in years and is no longer
useful in general; replace it with an off-by-default module param
to WARN if hardware fails a check that KVM does not perform
- Fix a currently-benign bug where KVM would drop the guest's
SPEC_CTRL[63:32] on VM-Enter
- Misc cleanups
- Overhaul the TDX code to address systemic races where KVM (acting
on behalf of userspace) could inadvertantly trigger lock contention
in the TDX-Module; KVM was either working around these in weird,
ugly ways, or was simply oblivious to them (though even Yan's
devilish selftests could only break individual VMs, not the host
kernel)
- Fix a bug where KVM could corrupt a vCPU's cpu_list when freeing a
TDX vCPU, if creating said vCPU failed partway through
- Fix a few sparse warnings (bad annotation, 0 != NULL)
- Use struct_size() to simplify copying TDX capabilities to userspace
- Fix a bug where TDX would effectively corrupt user-return MSR
values if the TDX Module rejects VP.ENTER and thus doesn't clobber
host MSRs as expected
Selftests:
- Fix a math goof in mmu_stress_test when running on a single-CPU
system/VM
- Forcefully override ARCH from x86_64 to x86 to play nice with
specifying ARCH=x86_64 on the command line
- Extend a bunch of nested VMX to validate nested SVM as well
- Add support for LA57 in the core VM_MODE_xxx macro, and add a test
to verify KVM can save/restore nested VMX state when L1 is using
5-level paging, but L2 is not
- Clean up the guest paging code in anticipation of sharing the core
logic for nested EPT and nested NPT
guest_memfd:
- Add NUMA mempolicy support for guest_memfd, and clean up a variety
of rough edges in guest_memfd along the way
- Define a CLASS to automatically handle get+put when grabbing a
guest_memfd from a memslot to make it harder to leak references
- Enhance KVM selftests to make it easer to develop and debug
selftests like those added for guest_memfd NUMA support, e.g. where
test and/or KVM bugs often result in hard-to-debug SIGBUS errors
- Misc cleanups
Generic:
- Use the recently-added WQ_PERCPU when creating the per-CPU
workqueue for irqfd cleanup
- Fix a goof in the dirty ring documentation
- Fix choice of target for directed yield across different calls to
kvm_vcpu_on_spin(); the function was always starting from the first
vCPU instead of continuing the round-robin search"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (260 commits)
KVM: arm64: at: Update AF on software walk only if VM has FEAT_HAFDBS
KVM: arm64: at: Use correct HA bit in TCR_EL2 when regime is EL2
KVM: arm64: Document KVM_PGTABLE_PROT_{UX,PX}
KVM: arm64: Fix spelling mistake "Unexpeced" -> "Unexpected"
KVM: arm64: Add break to default case in kvm_pgtable_stage2_pte_prot()
KVM: arm64: Add endian casting to kvm_swap_s[12]_desc()
KVM: arm64: Fix compilation when CONFIG_ARM64_USE_LSE_ATOMICS=n
KVM: arm64: selftests: Add test for AT emulation
KVM: arm64: nv: Expose hardware access flag management to NV guests
KVM: arm64: nv: Implement HW access flag management in stage-2 SW PTW
KVM: arm64: Implement HW access flag management in stage-1 SW PTW
KVM: arm64: Propagate PTW errors up to AT emulation
KVM: arm64: Add helper for swapping guest descriptor
KVM: arm64: nv: Use pgtable definitions in stage-2 walk
KVM: arm64: Handle endianness in read helper for emulated PTW
KVM: arm64: nv: Stop passing vCPU through void ptr in S2 PTW
KVM: arm64: Call helper for reading descriptors directly
KVM: arm64: nv: Advertise support for FEAT_XNX
KVM: arm64: Teach ptdump about FEAT_XNX permissions
KVM: s390: Use generic VIRT_XFER_TO_GUEST_WORK functions
...
In f2fs retry logic, we will call f2fs_io_schedule_timeout() to sleep as
uninterruptible state (waiting for IO) for a while, however, in several
paths below, we are not blocked by IO:
- f2fs_write_single_data_page() return -EAGAIN due to racing on cp_rwsem.
- f2fs_flush_device_cache() failed to submit preflush command.
- __issue_discard_cmd_range() sleeps periodically in between two in batch
discard submissions.
So, in order to reveal state of task more accurate, let's introduce
f2fs_schedule_timeout() and call it in above paths in where we are waiting
for non-IO reasons.
Then we can get real reason of uninterruptible sleep for a thread in
tracepoint, perfetto, etc.
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Change the type of the unlock parameter of f2fs_put_page to bool.
All callers should consistently pass true or false. No logical change.
Signed-off-by: Yongpeng Yang <yangyongpeng@xiaomi.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
This is one instruction more efficient than open-coding folio_pos() +
folio_size(). It's the equivalent of (x + y) << z rather than
x << z + y << z.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Link: https://patch.msgid.link/20251024170822.1427218-6-willy@infradead.org
Reviewed-by: Chao Yu <chao@kernel.org>
Cc: Jaegeuk Kim <jaegeuk@kernel.org>
Cc: Chao Yu <chao@kernel.org>
Cc: linux-f2fs-devel@lists.sourceforge.net
Signed-off-by: Christian Brauner <brauner@kernel.org>
Add a mempolicy parameter to filemap_alloc_folio() to enable NUMA-aware
page cache allocations. This will be used by upcoming changes to
support NUMA policies in guest-memfd, where guest_memory need to be
allocated NUMA policy specified by VMM.
All existing users pass NULL maintaining current behavior.
Reviewed-by: Pankaj Gupta <pankaj.gupta@amd.com>
Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Shivank Garg <shivankg@amd.com>
Tested-by: Ashish Kalra <ashish.kalra@amd.com>
Link: https://lore.kernel.org/r/20250827175247.83322-4-shivankg@amd.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
generic/091 may fail, then it bisects to the bad commit ba8dac350faf
("f2fs: fix to zero post-eof page").
What will cause generic/091 to fail is something like below Testcase #1:
1. write 16k as compressed blocks
2. truncate to 12k
3. truncate to 20k
4. verify data in range of [12k, 16k], however data is not zero as
expected
Script of Testcase #1
mkfs.f2fs -f -O extra_attr,compression /dev/vdb
mount -t f2fs -o compress_extension=* /dev/vdb /mnt/f2fs
dd if=/dev/zero of=/mnt/f2fs/file bs=12k count=1
dd if=/dev/random of=/mnt/f2fs/file bs=4k count=1 seek=3 conv=notrunc
sync
truncate -s $((12*1024)) /mnt/f2fs/file
truncate -s $((20*1024)) /mnt/f2fs/file
dd if=/mnt/f2fs/file of=/mnt/f2fs/data bs=4k count=1 skip=3
od /mnt/f2fs/data
umount /mnt/f2fs
Analisys:
in step 2), we will redirty all data pages from #0 to #3 in compressed
cluster, and zero page #3,
in step 3), f2fs_setattr() will call f2fs_zero_post_eof_page() to drop
all page cache post eof, includeing dirtied page #3,
in step 4) when we read data from page #3, it will decompressed cluster
and extra random data to page #3, finally, we hit the non-zeroed data
post eof.
However, the commit ba8dac350faf ("f2fs: fix to zero post-eof page") just
let the issue be reproduced easily, w/o the commit, it can reproduce this
bug w/ below Testcase #2:
1. write 16k as compressed blocks
2. truncate to 8k
3. truncate to 12k
4. truncate to 20k
5. verify data in range of [12k, 16k], however data is not zero as
expected
Script of Testcase #2
mkfs.f2fs -f -O extra_attr,compression /dev/vdb
mount -t f2fs -o compress_extension=* /dev/vdb /mnt/f2fs
dd if=/dev/zero of=/mnt/f2fs/file bs=12k count=1
dd if=/dev/random of=/mnt/f2fs/file bs=4k count=1 seek=3 conv=notrunc
sync
truncate -s $((8*1024)) /mnt/f2fs/file
truncate -s $((12*1024)) /mnt/f2fs/file
truncate -s $((20*1024)) /mnt/f2fs/file
echo 3 > /proc/sys/vm/drop_caches
dd if=/mnt/f2fs/file of=/mnt/f2fs/data bs=4k count=1 skip=3
od /mnt/f2fs/data
umount /mnt/f2fs
Anlysis:
in step 2), we will redirty all data pages from #0 to #3 in compressed
cluster, and zero page #2 and #3,
in step 3), we will truncate page #3 in page cache,
in step 4), expand file size,
in step 5), hit random data post eof w/ the same reason in Testcase #1.
Root Cause:
In f2fs_truncate_partial_cluster(), after we truncate partial data block
on compressed cluster, all pages in cluster including the one post eof
will be dirtied, after another tuncation, dirty page post eof will be
dropped, however on-disk compressed cluster is still valid, it may
include non-zero data post eof, result in exposing previous non-zero data
post eof while reading.
Fix:
In f2fs_truncate_partial_cluster(), let change as below to fix:
- call filemap_write_and_wait_range() to flush dirty page
- call truncate_pagecache() to drop pages or zero partial page post eof
- call f2fs_do_truncate_blocks() to truncate non-compress cluster to
last valid block
Fixes: 3265d3db1f16 ("f2fs: support partial truncation on compressed inode")
Reported-by: Jan Prusakowski <jprusakowski@google.com>
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Should cast type of folio->index from pgoff_t to loff_t to avoid overflow
while left shift operation.
Fixes: 3265d3db1f16 ("f2fs: support partial truncation on compressed inode")
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
The only caller already has a folio so pass it in.
f2fs_cache_compressed_page() is not used outside compress.c so
make it static. This requires a forward declaration (or would require
rearranging this file, but I've chosen not to do that for readability of
the diff).
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
All callers now have a folio so pass it in. Also remove the test for
the private flag; it is redundant with checking folio->private for being
NULL.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Both callers now have a folio so pass it in.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
All callers now have a folio so pass it in. Removes a call to
compound_head().
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
The only caller has a folio so pass it in.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
The only caller already has a folio so convert this function to be folio
based.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
The only caller has a folio, so pass it in and operate on it.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Remove a call to compound_head() by replacing a call to unlock_page()
with a call to folio_unlock().
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
The decompress_io_ctx may be released asynchronously after
I/O completion. If this file is deleted immediately after read,
and the kworker of processing post_read_wq has not been executed yet
due to high workloads, It is possible that the inode(f2fs_inode_info)
is evicted and freed before it is used f2fs_free_dic.
The UAF case as below:
Thread A Thread B
- f2fs_decompress_end_io
- f2fs_put_dic
- queue_work
add free_dic work to post_read_wq
- do_unlink
- iput
- evict
- call_rcu
This file is deleted after read.
Thread C kworker to process post_read_wq
- rcu_do_batch
- f2fs_free_inode
- kmem_cache_free
inode is freed by rcu
- process_scheduled_works
- f2fs_late_free_dic
- f2fs_free_dic
- f2fs_release_decomp_mem
read (dic->inode)->i_compress_algorithm
This patch store compress_algorithm and sbi in dic to avoid inode UAF.
In addition, the previous solution is deprecated in [1] may cause system hang.
[1] https://lore.kernel.org/all/c36ab955-c8db-4a8b-a9d0-f07b5f426c3f@kernel.org
Cc: Daeho Jeong <daehojeong@google.com>
Fixes: bff139b49d9f ("f2fs: handle decompress only post processing in softirq")
Signed-off-by: Zhiguo Niu <zhiguo.niu@unisoc.com>
Signed-off-by: Baocong Liu <baocong.liu@unisoc.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
No logic changes, just cleanup and prepare for fixing the UAF issue
in f2fs_free_dic.
Signed-off-by: Zhiguo Niu <zhiguo.niu@unisoc.com>
Signed-off-by: Baocong Liu <baocong.liu@unisoc.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Introduce a new fault type FAULT_VMALLOC to simulate no memory error in
f2fs_vmalloc().
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
.init_{,de}compress_ctx uses kvmalloc() to alloc memory, it will try
to allocate physically continuous page first, it may cause more memory
allocation pressure, let's use vmalloc instead to mitigate it.
[Test]
cd /data/local/tmp
touch file
f2fs_io setflags compression file
f2fs_io getflags file
for i in $(seq 1 10); do sync; echo 3 > /proc/sys/vm/drop_caches;\
time f2fs_io write 512 0 4096 zero osync file; truncate -s 0 file;\
done
[Result]
Before After Delta
21.243 21.694 -2.12%
For compression, we recommend to use ioctl to compress file data in
background for workaround.
For decompression, only zstd will be affected.
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Since __f2fs_crc32() now calls crc32() directly, it no longer uses its
sbi argument. Remove that, and simplify its callers accordingly.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Consolidate the code to unlock the page in f2fs_write_single_data_page
instead of leaving it to the callers for the AOP_WRITEPAGE_ACTIVATE case.
Replace AOP_WRITEPAGE_ACTIVATE with a positive return of 1 as this case
now doesn't match the historic ->writepage special return code that is
on it's way out now that ->writepage has been removed.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
All assignments to this struct member are conversions from a folio
so convert it to be a folio and convert all users. At the same time,
convert data_blkaddr() to take a folio as all callers now have a folio.
Remove eight calls to compound_head().
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
The only caller already has a folio, so pass it in. Copy the entire
size of the folio to support large block sizes. Remove two calls to
compound_head().
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Add f2fs_filemap_get_folio() as a wrapper around __filemap_get_folio()
which can inject an error. Removes seven calls to compound_head().
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Look up a folio instead of a page, and if that fails, allocate a folio.
Removes five calls to compound_head(), one of the last few references to
add_to_page_cache_lru() and honours the cpuset_do_page_mem_spread()
setting.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
The only caller has a folio, so pass it in.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Iterate over each folio rather than each page. Convert
f2fs_compress_control_page() to f2fs_compress_control_folio() since
this is the only caller. Removes a reference to page->mapping which
is going away soon as well as calls to fscrypt_is_bounce_page() and
fscrypt_pagecache_page().
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Remove a reference to page->mapping which is going away soon.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Convert each page in rpages to a folio before operating on it. Replaces
eight calls to compound_head() with one and removes a reference to
page->mapping which is going away soon.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Convert the incoming page to a folio and operate on it. Removes a
reference to page->mapping which is going away soon.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Jan Prusakowski reported a kernel hang issue as below:
When running xfstests on linux-next kernel (6.14.0-rc3, 6.12) I
encountered a problem in generic/475 test where fsstress process
gets blocked in __f2fs_write_data_pages() and the test hangs.
The options I used are:
MKFS_OPTIONS -- -O compression -O extra_attr -O project_quota -O quota /dev/vdc
MOUNT_OPTIONS -- -o acl,user_xattr -o discard,compress_extension=* /dev/vdc /vdc
INFO: task kworker/u8:0:11 blocked for more than 122 seconds.
Not tainted 6.14.0-rc3-xfstests-lockdep #1
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
task:kworker/u8:0 state:D stack:0 pid:11 tgid:11 ppid:2 task_flags:0x4208160 flags:0x00004000
Workqueue: writeback wb_workfn (flush-253:0)
Call Trace:
<TASK>
__schedule+0x309/0x8e0
schedule+0x3a/0x100
schedule_preempt_disabled+0x15/0x30
__mutex_lock+0x59a/0xdb0
__f2fs_write_data_pages+0x3ac/0x400
do_writepages+0xe8/0x290
__writeback_single_inode+0x5c/0x360
writeback_sb_inodes+0x22f/0x570
wb_writeback+0xb0/0x410
wb_do_writeback+0x47/0x2f0
wb_workfn+0x5a/0x1c0
process_one_work+0x223/0x5b0
worker_thread+0x1d5/0x3c0
kthread+0xfd/0x230
ret_from_fork+0x31/0x50
ret_from_fork_asm+0x1a/0x30
</TASK>
The root cause is: once generic/475 starts toload error table to dm
device, f2fs_prepare_compress_overwrite() will loop reading compressed
cluster pages due to IO error, meanwhile it has held .writepages lock,
it can block all other writeback tasks.
Let's fix this issue w/ below changes:
- add f2fs_handle_page_eio() in prepare_compress_overwrite() to
detect IO error.
- detect cp_error earler in f2fs_read_multi_pages().
Fixes: 4c8ff7095bef ("f2fs: support data compression")
Reported-by: Jan Prusakowski <jprusakowski@google.com>
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
When f2fs_write_single_data_page fails, f2fs_write_cache_pages
will use the last 'submitted' value incorrectly, which will cause
'nwritten' and 'wbc->nr_to_write' calculation errors
Signed-off-by: zangyangyang1 <zangyangyang1@xiaomi.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
New function can process some consecutive blocks at a time.
Function f2fs_invalidate_blocks()->down_write() and up_write()
are very time-consuming, so if f2fs_invalidate_blocks() can
process consecutive blocks at one time, it will save a lot of time.
Signed-off-by: Yi Sun <yi.sun@unisoc.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
New function f2fs_invalidate_compress_pages_range() adds the @len
parameter. So it can process some consecutive blocks at a time.
Signed-off-by: Yi Sun <yi.sun@unisoc.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Remove accesses to page->index and an unnecessary reference to
page->mapping.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Convert the incoming page to a folio and use it throughout.
Removes an access to page->index.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
This removes an access of page->index.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Remove references to page->index and use folio_test_uptodate()
instead of PageUptodate().
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Convert to use folio, so that we can get rid of 'page->index' to
prepare for removal of 'index' field in structure page [1].
[1] https://lore.kernel.org/all/Zp8fgUSIBGQ1TN0D@casper.infradead.org/
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Chao Yu <chao@kernel.org>
Reviewed-by: Li Zetao <lizetao1@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Convert to use folio, so that we can get rid of 'page->index' to
prepare for removal of 'index' field in structure page [1].
[1] https://lore.kernel.org/all/Zp8fgUSIBGQ1TN0D@casper.infradead.org/
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Chao Yu <chao@kernel.org>
Reviewed-by: Li Zetao <lizetao1@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Convert to use folio, so that we can get rid of 'page->index' to
prepare for removal of 'index' field in structure page [1].
[1] https://lore.kernel.org/all/Zp8fgUSIBGQ1TN0D@casper.infradead.org/
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
onvert to use folio, so that we can get rid of 'page->index' to
prepare for removal of 'index' field in structure page [1].
[1] https://lore.kernel.org/all/Zp8fgUSIBGQ1TN0D@casper.infradead.org/
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
In f2fs_do_write_data_page, when the data block is NULL_ADDR, it skips
writepage considering that it has been already truncated.
This results in an infinite loop as the PAGECACHE_TAG_TOWRITE tag is not
cleared during the writeback process for a compressed file including
NULL_ADDR in compress_mode=user.
This is the reproduction process:
1. dd if=/dev/zero bs=4096 count=1024 seek=1024 of=testfile
2. f2fs_io compress testfile
3. dd if=/dev/zero bs=4096 count=1 conv=notrunc of=testfile
4. f2fs_io decompress testfile
To prevent the problem, let's check whether the cluster is fully
allocated before redirty its pages.
Fixes: 5fdb322ff2c2 ("f2fs: add F2FS_IOC_DECOMPRESS_FILE and F2FS_IOC_COMPRESS_FILE")
Reviewed-by: Sungjong Seo <sj1557.seo@samsung.com>
Reviewed-by: Sunmin Jeong <s_min.jeong@samsung.com>
Tested-by: Jaewook Kim <jw5454.kim@samsung.com>
Signed-off-by: Yeongjin Gil <youngjin.gil@samsung.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Commit b1c9d3f833ba ("f2fs: support printk_ratelimited() in f2fs_printk()")
missed some cases, cover all remains for cleanup.
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
