December 2025 - Linux-stable-mirror

[PATCH 1/2] btrfs: fix beyond-EOF write handling

by Qu Wenruo

[BUG] For the following write sequence with 64K page size and 4K fs block size, it will lead to file extent items to be inserted without any data checksum: mkfs.btrfs -s 4k -f $dev > /dev/null mount $dev $mnt xfs_io -f -c "pwrite 0 16k" -c "pwrite 32k 4k" -c pwrite "60k 64K" \ -c "truncate 16k" $mnt/foobar umount $mnt This will result the following 2 file extent items to be inserted (extra trace point added to insert_ordered_extent_file_extent()): btrfs_finish_one_ordered: root=5 ino=257 file_off=61440 num_bytes=4096 csum_bytes=0 btrfs_finish_one_ordered: root=5 ino=257 file_off=0 num_bytes=16384 csum_bytes=16384 Note for file offset 60K, we're inserting an file extent without any data checksum. Also note that range [32K, 36K) didn't reach insert_ordered_extent_file_extent(), which is the correct behavior as that OE is fully truncated, should not result any file extent. Although file extent at 60K will be later dropped by btrfs_truncate(), if the transaction got committed after file extent inserted but before the file extent dropping, we will have a small window where we have a file extent beyond EOF and without any data checksum. That will cause "btrfs check" to report error. [CAUSE] The sequence happens like this: - Buffered write dirtied the page cache and updated isize Now the inode size is 64K, with the following page cache layout: 0 16K 32K 48K 64K |/////////////| |//| |//| - Truncate the inode to 16K Which will trigger writeback through: btrfs_setsize() |- truncate_setsize() | Now the inode size is set to 16K | |- btrfs_truncacte() |- btrfs_wait_ordered_range() for [16K, u64(-1)] |- btrfs_fdatawrite_range() for [16K, u64(-1)} |- extent_writepage() for folio 0 |- writepage_delalloc() | Generated OE for [0, 16K), [32K, 36K] and [60K, 64K) | |- extent_writepage_io() Then inside extent_writepage_io(), the dirty fs blocks are handled differently: - Submit write for range [0, 16K) As they are still inside the inode size (16K). - Mark OE [32K, 36K) as truncated Since we only call btrfs_lookup_first_ordered_range() once, which returned the first OE after file offset 16K. - Mark all OEs inside range [16K, 64K) as finished Which will mark OE ranges [32K, 36K) and [60K, 64K) as finished. For OE [32K, 36K) since it's already marked as truncated, and its truncated length is 0, no file extent will be inserted. For OE [60K, 64K) it has never been submitted thus has no data checksum, and we insert the file extent as usual. This is the root cause of file extent at 60K to be inserted without any data checksum. - Clear dirty flags for range [16K, 64K) It is the function btrfs_folio_clear_dirty() which search and clear any dirty blocks inside that range. [FIX] The bug itself is introduced a long time ago, way before subpage and larger folio support. At that time, fs block size must match page size, thus the range [cur, end) is just one fs block. But later with subpage and larger folios, the same range [cur, end) can have multiple blocks and ordered extents. Later commit 18de34daa7c6 ("btrfs: truncate ordered extent when skipping writeback past i_size") is fixing a bug related to subpage/larger folios, but it's still utilizing the old range [cur, end), meaning only the first OE will be marked as truncated. The proper fix here is to make EOF handling block-by-block, not trying to handle the whole range to @end. By this we always locate and truncate the OE for every dirty block. Cc: stable(a)vger.kernel.org # 5.15+ Signed-off-by: Qu Wenruo <wqu(a)suse.com> --- fs/btrfs/extent_io.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 629fd5af4286..a4b74023618d 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -1728,7 +1728,7 @@ static noinline_for_stack int extent_writepage_io(struct btrfs_inode *inode, struct btrfs_ordered_extent *ordered; ordered = btrfs_lookup_first_ordered_range(inode, cur, - folio_end - cur); + fs_info->sectorsize); /* * We have just run delalloc before getting here, so * there must be an ordered extent. @@ -1742,7 +1742,7 @@ static noinline_for_stack int extent_writepage_io(struct btrfs_inode *inode, btrfs_put_ordered_extent(ordered); btrfs_mark_ordered_io_finished(inode, folio, cur, - end - cur, true); + fs_info->sectorsize, true); /* * This range is beyond i_size, thus we don't need to * bother writing back. @@ -1751,8 +1751,8 @@ static noinline_for_stack int extent_writepage_io(struct btrfs_inode *inode, * writeback the sectors with subpage dirty bits, * causing writeback without ordered extent. */ - btrfs_folio_clear_dirty(fs_info, folio, cur, end - cur); - break; + btrfs_folio_clear_dirty(fs_info, folio, cur, fs_info->sectorsize); + continue; } ret = submit_one_sector(inode, folio, cur, bio_ctrl, i_size); if (unlikely(ret < 0)) { -- 2.52.0

2 weeks, 1 day

3
2
0 0

[PATCH v2] drm/nouveau/dispnv50: Don't call drm_atomic_get_crtc_state() in prepare_fb

by Lyude Paul

Since we recently started warning about uses of this function after the atomic check phase completes, we've started getting warnings about this in nouveau. It appears a misplaced drm_atomic_get_crtc_state() call has been hiding in our .prepare_fb callback for a while. So, fix this by adding a new nv50_head_atom_get_new() function and use that in our .prepare_fb callback instead. Signed-off-by: Lyude Paul <lyude(a)redhat.com> Fixes: 1590700d94ac ("drm/nouveau/kms/nv50-: split each resource type into their own source files") Cc: <stable(a)vger.kernel.org> # v4.18+ --- V2: * Don't call IS_ERR against the return value of drm_atomic_get_new_crtc_state(), it doesn't return pointer errors. Signed-off-by: Lyude Paul <lyude(a)redhat.com> --- drivers/gpu/drm/nouveau/dispnv50/atom.h | 13 +++++++++++++ drivers/gpu/drm/nouveau/dispnv50/wndw.c | 2 +- 2 files changed, 14 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/nouveau/dispnv50/atom.h b/drivers/gpu/drm/nouveau/dispnv50/atom.h index 93f8f4f645784..b43c4f9bbcdf5 100644 --- a/drivers/gpu/drm/nouveau/dispnv50/atom.h +++ b/drivers/gpu/drm/nouveau/dispnv50/atom.h @@ -152,8 +152,21 @@ static inline struct nv50_head_atom * nv50_head_atom_get(struct drm_atomic_state *state, struct drm_crtc *crtc) { struct drm_crtc_state *statec = drm_atomic_get_crtc_state(state, crtc); + if (IS_ERR(statec)) return (void *)statec; + + return nv50_head_atom(statec); +} + +static inline struct nv50_head_atom * +nv50_head_atom_get_new(struct drm_atomic_state *state, struct drm_crtc *crtc) +{ + struct drm_crtc_state *statec = drm_atomic_get_new_crtc_state(state, crtc); + + if (!statec) + return NULL; + return nv50_head_atom(statec); } diff --git a/drivers/gpu/drm/nouveau/dispnv50/wndw.c b/drivers/gpu/drm/nouveau/dispnv50/wndw.c index ef9e410babbfb..9a2c20fce0f3e 100644 --- a/drivers/gpu/drm/nouveau/dispnv50/wndw.c +++ b/drivers/gpu/drm/nouveau/dispnv50/wndw.c @@ -583,7 +583,7 @@ nv50_wndw_prepare_fb(struct drm_plane *plane, struct drm_plane_state *state) asyw->image.offset[0] = nvbo->offset; if (wndw->func->prepare) { - asyh = nv50_head_atom_get(asyw->state.state, asyw->state.crtc); + asyh = nv50_head_atom_get_new(asyw->state.state, asyw->state.crtc); if (IS_ERR(asyh)) return PTR_ERR(asyh); base-commit: c7685d11108acb387e44e3d81194d0d8959eaa44 -- 2.52.0

2 weeks, 1 day

1
0
0 0

[PATCH] KVM: x86/pmu: Do not accidentally create BTS events

by Fernand Sieber

From: Jan H. Schönherr <jschoenh(a)amazon.de> It is possible to degrade host performance by manipulating performance counters from a VM and tricking the host hypervisor to enable branch tracing. When the guest programs a CPU to track branch instructions and deliver an interrupt after exactly one branch instruction, the value one is handled by the host KVM/perf subsystems and treated incorrectly as a special value to enable the branch trace store (BTS) subsystem. It should not be possible to enable BTS from a guest. When BTS is enabled, it leads to general host performance degradation to both VMs and host. Perf considers the combination of PERF_COUNT_HW_BRANCH_INSTRUCTIONS with a sample_period of 1 a special case and handles this as a BTS event (see intel_pmu_has_bts_period()) -- a deviation from the usual semantic, where the sample_period represents the amount of branch instructions to encounter before the overflow handler is invoked. Nothing prevents a guest from programming its vPMU with the above settings (count branch, interrupt after one branch), which causes KVM to erroneously instruct perf to create a BTS event within pmc_reprogram_counter(), which does not have the desired semantics. The guest could also do more benign actions and request an interrupt after a more reasonable number of branch instructions via its vPMU. In that case counting works initially. However, KVM occasionally pauses and resumes the created performance counters. If the remaining amount of branch instructions until interrupt has reached 1 exactly, pmc_resume_counter() fails to resume the counter and a BTS event is created instead with its incorrect semantics. Fix this behavior by not passing the special value "1" as sample_period to perf. Instead, perform the same quirk that happens later in x86_perf_event_set_period() anyway, when the performance counter is transferred to the actual PMU: bump the sample_period to 2. Testing: From guest: `./wrmsr -p 12 0x186 0x1100c4` `./wrmsr -p 12 0xc1 0xffffffffffff` `./wrmsr -p 12 0x186 0x5100c4` This sequence sets up branch instruction counting, initializes the counter to overflow after one event (0xffffffffffff), and then enables edge detection (bit 18) for branch events. ./wrmsr -p 12 0x186 0x1100c4 Writes to IA32_PERFEVTSEL0 (0x186) Value 0x1100c4 breaks down as: Event = 0xC4 (Branch instructions) Bits 16-17: 0x1 (User mode only) Bit 22: 1 (Enable counter) ./wrmsr -p 12 0xc1 0xffffffffffff Writes to IA32_PMC0 (0xC1) Sets counter to maximum value (0xffffffffffff) This effectively sets up the counter to overflow on the next branch ./wrmsr -p 12 0x186 0x5100c4 Updates IA32_PERFEVTSEL0 again Similar to first command but adds bit 18 (0x4 to 0x5) Enables edge detection (bit 18) These MSR writes are trapped by the hypervisor in KVM and forwarded to the perf subsystem to create corresponding monitoring events. It is possible to repro this problem in a more realistic guest scenario: `perf record -e branches:u -c 2 -a &` `perf record -e branches:u -c 2 -a &` This presumably triggers the issue by KVM pausing and resuming the performance counter at the wrong moment, when its value is about to overflow. Signed-off-by: Jan H. Schönherr <jschoenh(a)amazon.de> Signed-off-by: Fernand Sieber <sieberf(a)amazon.com> Reviewed-by: David Woodhouse <dwmw(a)amazon.co.uk> Reviewed-by: Hendrik Borghorst <hborghor(a)amazon.de> Link: https://lore.kernel.org/r/20251124100220.238177-1-sieberf@amazon.com --- arch/x86/kvm/pmu.c | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/arch/x86/kvm/pmu.c b/arch/x86/kvm/pmu.c index 487ad19a236e..547512028e24 100644 --- a/arch/x86/kvm/pmu.c +++ b/arch/x86/kvm/pmu.c @@ -225,6 +225,19 @@ static u64 get_sample_period(struct kvm_pmc *pmc, u64 counter_value) { u64 sample_period = (-counter_value) & pmc_bitmask(pmc); + /* + * A sample_period of 1 might get mistaken by perf for a BTS event, see + * intel_pmu_has_bts_period(). This would prevent re-arming the counter + * via pmc_resume_counter(), followed by the accidental creation of an + * actual BTS event, which we do not want. + * + * Avoid this by bumping the sampling period. Note, that we do not lose + * any precision, because the same quirk happens later anyway (for + * different reasons) in x86_perf_event_set_period(). + */ + if (sample_period == 1) + sample_period = 2; + if (!sample_period) sample_period = pmc_bitmask(pmc) + 1; return sample_period; -- 2.43.0 Amazon Development Centre (South Africa) (Proprietary) Limited 29 Gogosoa Street, Observatory, Cape Town, Western Cape, 7925, South Africa Registration Number: 2004 / 034463 / 07

2 weeks, 1 day

5
11
0 0

[PATCH v2] fjes: Add missing iounmap in fjes_hw_init()

by Haoxiang Li

In error paths, add fjes_hw_iounmap() to release the resource acquired by fjes_hw_iomap(). Add a goto label to do so. Fixes: 8cdc3f6c5d22 ("fjes: Hardware initialization routine") Cc: stable(a)vger.kernel.org Signed-off-by: Haoxiang Li <lihaoxiang(a)isrc.iscas.ac.cn> Signed-off-by: Simon Horman <horms(a)kernel.org> --- Changes in v2: - Use an idiomatic goto to do the error hanlding. - Thanks for pointing out the issues with the patch, Simon! --- drivers/net/fjes/fjes_hw.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/drivers/net/fjes/fjes_hw.c b/drivers/net/fjes/fjes_hw.c index b9b5554ea862..5ad2673f213d 100644 --- a/drivers/net/fjes/fjes_hw.c +++ b/drivers/net/fjes/fjes_hw.c @@ -334,7 +334,7 @@ int fjes_hw_init(struct fjes_hw *hw) ret = fjes_hw_reset(hw); if (ret) - return ret; + goto err_iounmap; fjes_hw_set_irqmask(hw, REG_ICTL_MASK_ALL, true); @@ -347,8 +347,10 @@ int fjes_hw_init(struct fjes_hw *hw) hw->max_epid = fjes_hw_get_max_epid(hw); hw->my_epid = fjes_hw_get_my_epid(hw); - if ((hw->max_epid == 0) || (hw->my_epid >= hw->max_epid)) - return -ENXIO; + if ((hw->max_epid == 0) || (hw->my_epid >= hw->max_epid)) { + ret = -ENXIO; + goto err_iounmap; + } ret = fjes_hw_setup(hw); @@ -356,6 +358,10 @@ int fjes_hw_init(struct fjes_hw *hw) hw->hw_info.trace_size = FJES_DEBUG_BUFFER_SIZE; return ret; + +err_iounmap: + fjes_hw_iounmap(hw); + return ret; } void fjes_hw_exit(struct fjes_hw *hw) -- 2.25.1

2 weeks, 1 day

4
3
0 0

[PATCH v2] media: dvb-frontends: Fix out of bound for buffer

by Ilya Krutskih

Result of pointer shifting is out of bound for buffer 'prev_swp_freq' because of incrementation inside 'while' cycle in case more than 20 steps of it. Add check: prev_freq_num < 20 Cc: stable(a)vger.kernel.org # v5.10+ Fixes: 9a0bf528b4d6 ([media] move the dvb/frontends to drivers/media/dvb-frontends) Signed-off-by: Ilya Krutskih <devsec(a)tpz.ru> --- drivers/media/dvb-frontends/mb86a16.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/media/dvb-frontends/mb86a16.c b/drivers/media/dvb-frontends/mb86a16.c index 9033e39d75f4..da8985b91475 100644 --- a/drivers/media/dvb-frontends/mb86a16.c +++ b/drivers/media/dvb-frontends/mb86a16.c @@ -1188,7 +1188,7 @@ static int mb86a16_set_fe(struct mb86a16_state *state) signal_dupl = 0; for (j = 0; j < prev_freq_num; j++) { - if ((abs(prev_swp_freq[j] - swp_freq)) < (swp_ofs * 3 / 2)) { + if ((prev_freq_num < 20) && (abs(prev_swp_freq[j] - swp_freq)) < (swp_ofs * 3 / 2)) { signal_dupl = 1; dprintk(verbose, MB86A16_INFO, 1, "Probably Duplicate Signal, j = %d", j); } -- 2.43.0

2 weeks, 1 day

1
0
0 0

[PATCH v3] zstd: fixed possible 'rtbTable' underflow in FSE_normalizeCount()

by Ilya Krutskih

'rtbTable' may be underflowed because 'proba' is used without checking for a non-negative as index of rtbTable[]. Add check: proba >= 0 Cc: stable(a)vger.kernel.org # v5.10+ Fixes: e0c1b49f5b67 ("lib: zstd: Upgrade to latest upstream zstd version 1.4.10") Signed-off-by: Ilya Krutskih <devsec(a)tpz.ru> --- lib/zstd/compress/fse_compress.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/lib/zstd/compress/fse_compress.c b/lib/zstd/compress/fse_compress.c index 44a3c10becf2..6b83f8bc943a 100644 --- a/lib/zstd/compress/fse_compress.c +++ b/lib/zstd/compress/fse_compress.c @@ -492,9 +492,10 @@ size_t FSE_normalizeCount (short* normalizedCounter, unsigned tableLog, stillToDistribute--; } else { short proba = (short)((count[s]*step) >> scale); - if (proba<8) { - U64 restToBeat = vStep * rtbTable[proba]; - proba += (count[s]*step) - ((U64)proba<<scale) > restToBeat; + if ((proba >= 0) && (proba < 8)) { + U64 restToBeat = vStep * rtbTable[proba]; + + proba += (count[s]*step) - ((U64)proba<<scale) > restToBeat; } if (proba > largestP) { largestP=proba; largest=s; } normalizedCounter[s] = proba; -- 2.43.0

2 weeks, 1 day

1
0
0 0

[PATCH v4 01/22] drm/xe/svm: Fix a debug printout

by Thomas Hellström

Avoid spamming the log with drm_info(). Use drm_dbg() instead. Fixes: cc795e041034 ("drm/xe/svm: Make xe_svm_range_needs_migrate_to_vram() public") Cc: Matthew Brost <matthew.brost(a)intel.com> Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray(a)intel.com> Cc: <stable(a)vger.kernel.org> # v6.17+ Signed-off-by: Thomas Hellström <thomas.hellstrom(a)linux.intel.com> Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray(a)intel.com> --- drivers/gpu/drm/xe/xe_svm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/xe/xe_svm.c b/drivers/gpu/drm/xe/xe_svm.c index 46977ec1e0de..36634c84d148 100644 --- a/drivers/gpu/drm/xe/xe_svm.c +++ b/drivers/gpu/drm/xe/xe_svm.c @@ -948,7 +948,7 @@ bool xe_svm_range_needs_migrate_to_vram(struct xe_svm_range *range, struct xe_vm xe_assert(vm->xe, IS_DGFX(vm->xe)); if (xe_svm_range_in_vram(range)) { - drm_info(&vm->xe->drm, "Range is already in VRAM\n"); + drm_dbg(&vm->xe->drm, "Range is already in VRAM\n"); return false; } -- 2.51.1

2 weeks, 1 day

1
0
0 0

[PATCH v4] pmdomain: imx: Fix reference count leak in imx_gpc_probe()

by Wentao Liang

of_get_child_by_name() returns a node pointer with refcount incremented. Use the __free() attribute to manage the pgc_node reference, ensuring automatic of_node_put() cleanup when pgc_node goes out of scope. This eliminates the need for explicit error handling paths and avoids reference count leaks. Fixes: 721cabf6c660 ("soc: imx: move PGC handling to a new GPC driver") Cc: stable(a)vger.kernel.org Signed-off-by: Wentao Liang <vulab(a)iscas.ac.cn> --- Change in V4: - Fix typo error in code Change in V3: - Ensure variable is assigned when using cleanup attribute Change in V2: - Use __free() attribute instead of explicit of_node_put() calls --- drivers/pmdomain/imx/gpc.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/drivers/pmdomain/imx/gpc.c b/drivers/pmdomain/imx/gpc.c index f18c7e6e75dd..56a78cc86584 100644 --- a/drivers/pmdomain/imx/gpc.c +++ b/drivers/pmdomain/imx/gpc.c @@ -403,13 +403,12 @@ static int imx_gpc_old_dt_init(struct device *dev, struct regmap *regmap, static int imx_gpc_probe(struct platform_device *pdev) { const struct imx_gpc_dt_data *of_id_data = device_get_match_data(&pdev->dev); - struct device_node *pgc_node; + struct device_node *pgc_node __free(device_node) + = of_get_child_by_name(pdev->dev.of_node, "pgc"); struct regmap *regmap; void __iomem *base; int ret; - pgc_node = of_get_child_by_name(pdev->dev.of_node, "pgc"); - /* bail out if DT too old and doesn't provide the necessary info */ if (!of_property_present(pdev->dev.of_node, "#power-domain-cells") && !pgc_node) -- 2.34.1

2 weeks, 1 day

2
1
0 0

[PATCH v3] drivers/pci: Decouple DPC from AER service

by Darshit Shah

According to PCIe r7.0, sec. 6.2.11, "Implementation Note: Determination of DPC Control", it is recommended that the Operating System link the enablement of Downstream Port Containment (DPC) to the enablement of Advanced Error Reporting (AER). However, AER is advertised only on Root Port (RP) or Root Complex Event Collector (RCEC) devices. On the other hand, DPC may be advertised on any PCIe device in the hierarchy. In fact, since the main usecase of DPC is for the switch upstream of an endpoint device to trigger a signal to the host-bridge, it is imperative that it be supported on non-RP, non-RCEC devices. Previously portdrv has interpreted the spec to mean that the AER service must be available on the same device in order for DPC to be available. This is not what the implementation note meant to imply. If the firmware hands Linux control of AER via _OSC on the host bridge upstream of the device, then Linux should be allowed to assume control of DPC on the device. The comment above this check alludes to this, by saying: > With dpc-native, allow Linux to use DPC even if it doesn't have > permission to use AER. However, permission to use AER is negotiated at the host bridge, not per-device. So we should not link DPC to enabling AER at the device. Instead, DPC should be enabled if the OS has control of AER for the host bridge that is upstream of the device in question, or if dpc-native was set on the command line. Cc: stable(a)vger.kernel.org Signed-off-by: Darshit Shah <darnshah(a)amazon.de> Reviewed-by: Lukas Wunner <lukas(a)wunner.de> Reviewed-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy(a)linux.intel.com> --- Changes from v2: * + stable(a)vger.kernel.org since moving to v6.2+ breaks DPC on our systems * Minor stylistic changes to commit message * NO functional changes drivers/pci/pcie/portdrv.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/pci/pcie/portdrv.c b/drivers/pci/pcie/portdrv.c index 38a41ccf79b9..8db2fa140ae2 100644 --- a/drivers/pci/pcie/portdrv.c +++ b/drivers/pci/pcie/portdrv.c @@ -264,7 +264,7 @@ static int get_port_device_capability(struct pci_dev *dev) */ if (pci_find_ext_capability(dev, PCI_EXT_CAP_ID_DPC) && pci_aer_available() && - (pcie_ports_dpc_native || (services & PCIE_PORT_SERVICE_AER))) + (host->native_aer || pcie_ports_dpc_native)) services |= PCIE_PORT_SERVICE_DPC; /* Enable bandwidth control if more than one speed is supported. */ -- 2.47.3 Amazon Web Services Development Center Germany GmbH Tamara-Danz-Str. 13 10243 Berlin Geschaeftsfuehrung: Christof Hellmis, Andreas Stieger Eingetragen am Amtsgericht Charlottenburg unter HRB 257764 B Sitz: Berlin Ust-ID: DE 365 538 597

2 weeks, 1 day

1
0
0 0

[PATCH sched_ext/for-6.19-fixes] sched_ext: Fix bypass depth leak on scx_enable() failure

by Tejun Heo

scx_enable() calls scx_bypass(true) to initialize in bypass mode and then scx_bypass(false) on success to exit. If scx_enable() fails during task initialization - e.g. scx_cgroup_init() or scx_init_task() returns an error - it jumps to err_disable while bypass is still active. scx_disable_workfn() then calls scx_bypass(true/false) for its own bypass, leaving the bypass depth at 1 instead of 0. This causes the system to remain permanently in bypass mode after a failed scx_enable(). Failures after task initialization is complete - e.g. scx_tryset_enable_state() at the end - already call scx_bypass(false) before reaching the error path and are not affected. This only affects a subset of failure modes. Fix it by tracking whether scx_enable() called scx_bypass(true) in a bool and having scx_disable_workfn() call an extra scx_bypass(false) to clear it. This is a temporary measure as the bypass depth will be moved into the sched instance, which will make this tracking unnecessary. Fixes: 8c2090c504e9 ("sched_ext: Initialize in bypass mode") Cc: stable(a)vger.kernel.org # v6.12+ Reported-by: Chris Mason <clm(a)meta.com> Signed-off-by: Tejun Heo <tj(a)kernel.org> --- kernel/sched/ext.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) --- a/kernel/sched/ext.c +++ b/kernel/sched/ext.c @@ -41,6 +41,13 @@ static bool scx_init_task_enabled; static bool scx_switching_all; DEFINE_STATIC_KEY_FALSE(__scx_switched_all); +/* + * Tracks whether scx_enable() called scx_bypass(true). Used to balance bypass + * depth on enable failure. Will be removed when bypass depth is moved into the + * sched instance. + */ +static bool scx_bypassed_for_enable; + static atomic_long_t scx_nr_rejected = ATOMIC_LONG_INIT(0); static atomic_long_t scx_hotplug_seq = ATOMIC_LONG_INIT(0); @@ -4318,6 +4325,11 @@ static void scx_disable_workfn(struct kt scx_dsp_max_batch = 0; free_kick_syncs(); + if (scx_bypassed_for_enable) { + scx_bypassed_for_enable = false; + scx_bypass(false); + } + mutex_unlock(&scx_enable_mutex); WARN_ON_ONCE(scx_set_enable_state(SCX_DISABLED) != SCX_DISABLING); @@ -4970,6 +4982,7 @@ static int scx_enable(struct sched_ext_o * Init in bypass mode to guarantee forward progress. */ scx_bypass(true); + scx_bypassed_for_enable = true; for (i = SCX_OPI_NORMAL_BEGIN; i < SCX_OPI_NORMAL_END; i++) if (((void (**)(void))ops)[i]) @@ -5067,6 +5080,7 @@ static int scx_enable(struct sched_ext_o scx_task_iter_stop(&sti); percpu_up_write(&scx_fork_rwsem); + scx_bypassed_for_enable = false; scx_bypass(false); if (!scx_tryset_enable_state(SCX_ENABLED, SCX_ENABLING)) { -- tejun

2 weeks, 1 day

4
4
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror December 2025