August 2024 - Linux-stable-mirror

+ mm-numa-no-task_numa_fault-call-if-page-table-is-changed.patch added to mm-hotfixes-unstable branch

by Andrew Morton

The patch titled Subject: mm/numa: no task_numa_fault() call if page table is changed has been added to the -mm mm-hotfixes-unstable branch. Its filename is mm-numa-no-task_numa_fault-call-if-page-table-is-changed.patch This patch will shortly appear at https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patche… This patch will later appear in the mm-hotfixes-unstable branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/process/submit-checklist.rst when testing your code *** The -mm tree is included into linux-next via the mm-everything branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm and is updated there every 2-3 working days ------------------------------------------------------ From: Zi Yan <ziy(a)nvidia.com> Subject: mm/numa: no task_numa_fault() call if page table is changed Date: Wed, 7 Aug 2024 14:47:29 -0400 When handling a numa page fault, task_numa_fault() should be called by a process that restores the page table of the faulted folio to avoid duplicated stats counting. Commit b99a342d4f11 ("NUMA balancing: reduce TLB flush via delaying mapping on hint page fault") restructured do_numa_page() and do_huge_pmd_numa_page() and did not avoid task_numa_fault() call in the second page table check after a numa migration failure. Fix it by making all !pte_same()/!pmd_same() return immediately. This issue can cause task_numa_fault() being called more than necessary and lead to unexpected numa balancing results (It is hard to tell whether the issue will cause positive or negative performance impact due to duplicated numa fault counting). Link: https://lkml.kernel.org/r/20240807184730.1266736-1-ziy@nvidia.com Fixes: b99a342d4f11 ("NUMA balancing: reduce TLB flush via delaying mapping on hint page fault") Signed-off-by: Zi Yan <ziy(a)nvidia.com> Reported-by: "Huang, Ying" <ying.huang(a)intel.com> Closes: https://lore.kernel.org/linux-mm/87zfqfw0yw.fsf@yhuang6-desk2.ccr.corp.inte… Cc: Baolin Wang <baolin.wang(a)linux.alibaba.com> Cc: David Hildenbrand <david(a)redhat.com> Cc: Kefeng Wang <wangkefeng.wang(a)huawei.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- mm/huge_memory.c | 5 +++-- mm/memory.c | 5 +++-- 2 files changed, 6 insertions(+), 4 deletions(-) --- a/mm/huge_memory.c~mm-numa-no-task_numa_fault-call-if-page-table-is-changed +++ a/mm/huge_memory.c @@ -1738,10 +1738,11 @@ vm_fault_t do_huge_pmd_numa_page(struct goto out_map; } -out: +count_fault: if (nid != NUMA_NO_NODE) task_numa_fault(last_cpupid, nid, HPAGE_PMD_NR, flags); +out: return 0; out_map: @@ -1753,7 +1754,7 @@ out_map: set_pmd_at(vma->vm_mm, haddr, vmf->pmd, pmd); update_mmu_cache_pmd(vma, vmf->address, vmf->pmd); spin_unlock(vmf->ptl); - goto out; + goto count_fault; } /* --- a/mm/memory.c~mm-numa-no-task_numa_fault-call-if-page-table-is-changed +++ a/mm/memory.c @@ -5371,9 +5371,10 @@ static vm_fault_t do_numa_page(struct vm goto out_map; } -out: +count_fault: if (nid != NUMA_NO_NODE) task_numa_fault(last_cpupid, nid, nr_pages, flags); +out: return 0; out_map: /* @@ -5387,7 +5388,7 @@ out_map: numa_rebuild_single_mapping(vmf, vma, vmf->address, vmf->pte, writable); pte_unmap_unlock(vmf->pte, vmf->ptl); - goto out; + goto count_fault; } static inline vm_fault_t create_huge_pmd(struct vm_fault *vmf) _ Patches currently in -mm which might be from ziy(a)nvidia.com are mm-numa-no-task_numa_fault-call-if-page-table-is-changed.patch memory-tiering-read-last_cpupid-correctly-in-do_huge_pmd_numa_page.patch memory-tiering-introduce-folio_use_access_time-check.patch memory-tiering-count-pgpromote_success-when-mem-tiering-is-enabled.patch mm-migrate-move-common-code-to-numa_migrate_check-was-numa_migrate_prep.patch

4 months, 2 weeks

1
0
0 0

[PATCH 1/2] mseal: Fix is_madv_discard()

by Pedro Falcato

is_madv_discard did its check wrong. MADV_ flags are not bitwise, they're normal sequential numbers. So, for instance: behavior & (/* ... */ | MADV_REMOVE) tagged both MADV_REMOVE and MADV_RANDOM (bit 0 set) as discard operations. This is obviously incorrect, so use a switch statement instead. Cc: stable(a)vger.kernel.org Fixes: 8be7258aad44 ("mseal: add mseal syscall") Signed-off-by: Pedro Falcato <pedro.falcato(a)gmail.com> --- mm/mseal.c | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-) diff --git a/mm/mseal.c b/mm/mseal.c index bf783bba8ed..15bba28acc0 100644 --- a/mm/mseal.c +++ b/mm/mseal.c @@ -40,9 +40,17 @@ static bool can_modify_vma(struct vm_area_struct *vma) static bool is_madv_discard(int behavior) { - return behavior & - (MADV_FREE | MADV_DONTNEED | MADV_DONTNEED_LOCKED | - MADV_REMOVE | MADV_DONTFORK | MADV_WIPEONFORK); + switch (behavior) { + case MADV_FREE: + case MADV_DONTNEED: + case MADV_DONTNEED_LOCKED: + case MADV_REMOVE: + case MADV_DONTFORK: + case MADV_WIPEONFORK: + return true; + } + + return false; } static bool is_ro_anon(struct vm_area_struct *vma) -- 2.46.0

4 months, 2 weeks

2
3
0 0

+ lib-stackdepot-double-depot_pools_cap-if-kasan-is-enabled.patch added to mm-hotfixes-unstable branch

by Andrew Morton

The patch titled Subject: lib/stackdepot: double DEPOT_POOLS_CAP if KASAN is enabled has been added to the -mm mm-hotfixes-unstable branch. Its filename is lib-stackdepot-double-depot_pools_cap-if-kasan-is-enabled.patch This patch will shortly appear at https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patche… This patch will later appear in the mm-hotfixes-unstable branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/process/submit-checklist.rst when testing your code *** The -mm tree is included into linux-next via the mm-everything branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm and is updated there every 2-3 working days ------------------------------------------------------ From: Waiman Long <longman(a)redhat.com> Subject: lib/stackdepot: double DEPOT_POOLS_CAP if KASAN is enabled Date: Wed, 7 Aug 2024 12:52:28 -0400 When a wide variety of workloads are run on a debug kernel with KASAN enabled, the following warning may sometimes be printed. [ 6818.650674] Stack depot reached limit capacity [ 6818.650730] WARNING: CPU: 1 PID: 272741 at lib/stackdepot.c:252 depot_alloc_stack+0x39e/0x3d0 : [ 6818.650907] Call Trace: [ 6818.650909] [<00047dd453d84b92>] depot_alloc_stack+0x3a2/0x3d0 [ 6818.650916] [<00047dd453d85254>] stack_depot_save_flags+0x4f4/0x5c0 [ 6818.650920] [<00047dd4535872c6>] kasan_save_stack+0x56/0x70 [ 6818.650924] [<00047dd453587328>] kasan_save_track+0x28/0x40 [ 6818.650927] [<00047dd45358a27a>] kasan_save_free_info+0x4a/0x70 [ 6818.650930] [<00047dd45358766a>] __kasan_slab_free+0x12a/0x1d0 [ 6818.650933] [<00047dd45350deb4>] kmem_cache_free+0x1b4/0x580 [ 6818.650938] [<00047dd452c520da>] __put_task_struct+0x24a/0x320 [ 6818.650945] [<00047dd452c6aee4>] delayed_put_task_struct+0x294/0x350 [ 6818.650949] [<00047dd452e9066a>] rcu_do_batch+0x6ea/0x2090 [ 6818.650953] [<00047dd452ea60f4>] rcu_core+0x474/0xa90 [ 6818.650956] [<00047dd452c780c0>] handle_softirqs+0x3c0/0xf90 [ 6818.650960] [<00047dd452c76fbe>] __irq_exit_rcu+0x35e/0x460 [ 6818.650963] [<00047dd452c79992>] irq_exit_rcu+0x22/0xb0 [ 6818.650966] [<00047dd454bd8128>] do_ext_irq+0xd8/0x120 [ 6818.650972] [<00047dd454c0ddd0>] ext_int_handler+0xb8/0xe8 [ 6818.650979] [<00047dd453589cf6>] kasan_check_range+0x236/0x2f0 [ 6818.650982] [<00047dd453378cf0>] filemap_get_pages+0x190/0xaa0 [ 6818.650986] [<00047dd453379940>] filemap_read+0x340/0xa70 [ 6818.650989] [<00047dd3d325d226>] xfs_file_buffered_read+0x2c6/0x400 [xfs] [ 6818.651431] [<00047dd3d325dfe2>] xfs_file_read_iter+0x2c2/0x550 [xfs] [ 6818.651663] [<00047dd45364710c>] vfs_read+0x64c/0x8c0 [ 6818.651669] [<00047dd453648ed8>] ksys_read+0x118/0x200 [ 6818.651672] [<00047dd452b6cf5a>] do_syscall+0x27a/0x380 [ 6818.651676] [<00047dd454bd7e74>] __do_syscall+0xf4/0x1a0 [ 6818.651680] [<00047dd454c0db58>] system_call+0x70/0x98 As KASAN is a big user of stackdepot, the current DEPOT_POOLS_CAP of 8192 may not be enough. Double DEPOT_POOLS_CAP if KASAN is enabled to avoid hitting this problem. Also use the MIN() macro for defining DEPOT_MAX_POOLS to clarify the intention. Link: https://lkml.kernel.org/r/20240807165228.1116831-1-longman@redhat.com Fixes: 02754e0a484a ("lib/stackdepot.c: bump stackdepot capacity from 16MB to 128MB") Signed-off-by: Waiman Long <longman(a)redhat.com> Cc: Andrey Konovalov <andreyknvl(a)google.com> Cc: Andrey Ryabinin <ryabinin.a.a(a)gmail.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- lib/stackdepot.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) --- a/lib/stackdepot.c~lib-stackdepot-double-depot_pools_cap-if-kasan-is-enabled +++ a/lib/stackdepot.c @@ -36,11 +36,12 @@ #include <linux/memblock.h> #include <linux/kasan-enabled.h> -#define DEPOT_POOLS_CAP 8192 +/* KASAN is a big user of stackdepot, double the cap if KASAN is enabled */ +#define DEPOT_POOLS_CAP (8192 * (IS_ENABLED(CONFIG_KASAN) ? 2 : 1)) + /* The pool_index is offset by 1 so the first record does not have a 0 handle. */ #define DEPOT_MAX_POOLS \ - (((1LL << (DEPOT_POOL_INDEX_BITS)) - 1 < DEPOT_POOLS_CAP) ? \ - (1LL << (DEPOT_POOL_INDEX_BITS)) - 1 : DEPOT_POOLS_CAP) + MIN((1LL << (DEPOT_POOL_INDEX_BITS)) - 1, DEPOT_POOLS_CAP) static bool stack_depot_disabled; static bool __stack_depot_early_init_requested __initdata = IS_ENABLED(CONFIG_STACKDEPOT_ALWAYS_INIT); _ Patches currently in -mm which might be from longman(a)redhat.com are padata-fix-possible-divide-by-0-panic-in-padata_mt_helper.patch mm-memory-failure-use-raw_spinlock_t-in-struct-memory_failure_cpu.patch mm-memory-failure-use-raw_spinlock_t-in-struct-memory_failure_cpu-v3.patch lib-stackdepot-double-depot_pools_cap-if-kasan-is-enabled.patch watchdog-handle-the-enodev-failure-case-of-lockup_detector_delay_init-separately.patch

4 months, 2 weeks

1
0
0 0

FAILED: patch "[PATCH] selftests: mptcp: fix error path" failed to apply to 6.6-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.6-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. To reproduce the conflict and resubmit, you may use the following commands: git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-6.6.y git checkout FETCH_HEAD git cherry-pick -x 4a2f48992ddf4b8c2fba846c6754089edae6db5a # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2024080739-imperial-modular-7da5@gregkh' --subject-prefix 'PATCH 6.6.y' HEAD^.. Possible dependencies: 4a2f48992ddf ("selftests: mptcp: fix error path") 571d79664a4a ("selftests: mptcp: join: update endpoint ops") 0d16ed0c2e74 ("selftests: mptcp: add {get,format}_endpoint(s) helpers") 3188309c8ceb ("selftests: mptcp: netlink: add 'limits' helpers") 29aa32fee7d0 ("selftests: mptcp: export ip_mptcp to mptcp_lib") 40061817d95b ("selftests: mptcp: join: fix dev in check_endpoint") 7f0782ca1ce9 ("selftests: mptcp: add mptcp_lib_verify_listener_events") 8ebb44196585 ("selftests: mptcp: print_test out of verify_listener_events") 663260e14668 ("selftests: mptcp: extract mptcp_lib_check_expected") 339c225e2e03 ("selftests: mptcp: call test_fail without argument") 747ba8783a33 ("selftests: mptcp: print test results with colors") e7c42bf4d320 ("selftests: mptcp: use += operator to append strings") aa7694766f14 ("selftests: mptcp: print test results with counters") 3382bb09701b ("selftests: mptcp: add print_title in mptcp_lib") 9e6a39ecb9a1 ("selftests: mptcp: export TEST_COUNTER variable") fd959262c1bb ("selftests: mptcp: sockopt: print every test result") c9161a0f8ff9 ("selftests: mptcp: connect: fix misaligned output") 01ed9838107f ("selftests: mptcp: connect: add dedicated port counter") 6215df11b945 ("selftests: mptcp: print all error messages to stdout") 2aebd3579d90 ("selftests: mptcp: simult flows: fix shellcheck warnings") thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 4a2f48992ddf4b8c2fba846c6754089edae6db5a Mon Sep 17 00:00:00 2001 From: Paolo Abeni <pabeni(a)redhat.com> Date: Sat, 27 Jul 2024 11:04:02 +0200 Subject: [PATCH] selftests: mptcp: fix error path pm_nl_check_endpoint() currently calls an not existing helper to mark the test as failed. Fix the wrong call. Fixes: 03668c65d153 ("selftests: mptcp: join: rework detailed report") Cc: stable(a)vger.kernel.org Signed-off-by: Paolo Abeni <pabeni(a)redhat.com> Reviewed-by: Matthieu Baerts (NGI0) <matttbe(a)kernel.org> Signed-off-by: Matthieu Baerts (NGI0) <matttbe(a)kernel.org> Signed-off-by: David S. Miller <davem(a)davemloft.net> diff --git a/tools/testing/selftests/net/mptcp/mptcp_join.sh b/tools/testing/selftests/net/mptcp/mptcp_join.sh index 9c091fc267c4..55d84a1bde15 100755 --- a/tools/testing/selftests/net/mptcp/mptcp_join.sh +++ b/tools/testing/selftests/net/mptcp/mptcp_join.sh @@ -661,7 +661,7 @@ pm_nl_check_endpoint() done if [ -z "${id}" ]; then - test_fail "bad test - missing endpoint id" + fail_test "bad test - missing endpoint id" return fi

4 months, 2 weeks

2
1
0 0

FAILED: patch "[PATCH] mptcp: pm: fix backup support in signal endpoints" failed to apply to 6.6-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.6-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. To reproduce the conflict and resubmit, you may use the following commands: git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-6.6.y git checkout FETCH_HEAD git cherry-pick -x 6834097fc38c5416701c793da94558cea49c0a1f # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2024080726-educator-stubble-a08d@gregkh' --subject-prefix 'PATCH 6.6.y' HEAD^.. Possible dependencies: 6834097fc38c ("mptcp: pm: fix backup support in signal endpoints") 9ae7846c4b6b ("mptcp: dump addrs in userspace pm list") 34e74a5cf3b7 ("mptcp: implement mptcp_userspace_pm_dump_addr") aab4d8564947 ("net: mptcp: use policy generated by YAML spec") 1e07938e29c5 ("net: mptcp: rename netlink handlers to mptcp_pm_nl_<blah>_{doit,dumpit}") 1d0507f46843 ("net: mptcp: convert netlink from small_ops to ops") thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 6834097fc38c5416701c793da94558cea49c0a1f Mon Sep 17 00:00:00 2001 From: "Matthieu Baerts (NGI0)" <matttbe(a)kernel.org> Date: Sat, 27 Jul 2024 12:01:28 +0200 Subject: [PATCH] mptcp: pm: fix backup support in signal endpoints There was a support for signal endpoints, but only when the endpoint's flag was changed during a connection. If an endpoint with the signal and backup was already present, the MP_JOIN reply was not containing the backup flag as expected. That's confusing to have this inconsistent behaviour. On the other hand, the infrastructure to set the backup flag in the SYN + ACK + MP_JOIN was already there, it was just never set before. Now when requesting the local ID from the path-manager, the backup status is also requested. Note that when the userspace PM is used, the backup flag can be set if the local address was already used before with a backup flag, e.g. if the address was announced with the 'backup' flag, or a subflow was created with the 'backup' flag. Fixes: 4596a2c1b7f5 ("mptcp: allow creating non-backup subflows") Cc: stable(a)vger.kernel.org Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/507 Reviewed-by: Mat Martineau <martineau(a)kernel.org> Signed-off-by: Matthieu Baerts (NGI0) <matttbe(a)kernel.org> Signed-off-by: Paolo Abeni <pabeni(a)redhat.com> diff --git a/net/mptcp/pm.c b/net/mptcp/pm.c index 55406720c607..23bb89c94e90 100644 --- a/net/mptcp/pm.c +++ b/net/mptcp/pm.c @@ -426,6 +426,18 @@ int mptcp_pm_get_local_id(struct mptcp_sock *msk, struct sock_common *skc) return mptcp_pm_nl_get_local_id(msk, &skc_local); } +bool mptcp_pm_is_backup(struct mptcp_sock *msk, struct sock_common *skc) +{ + struct mptcp_addr_info skc_local; + + mptcp_local_address((struct sock_common *)skc, &skc_local); + + if (mptcp_pm_is_userspace(msk)) + return mptcp_userspace_pm_is_backup(msk, &skc_local); + + return mptcp_pm_nl_is_backup(msk, &skc_local); +} + int mptcp_pm_get_flags_and_ifindex_by_id(struct mptcp_sock *msk, unsigned int id, u8 *flags, int *ifindex) { diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c index 7635fac91539..37954a0b087d 100644 --- a/net/mptcp/pm_netlink.c +++ b/net/mptcp/pm_netlink.c @@ -1101,6 +1101,24 @@ int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk, struct mptcp_addr_info *skc return ret; } +bool mptcp_pm_nl_is_backup(struct mptcp_sock *msk, struct mptcp_addr_info *skc) +{ + struct pm_nl_pernet *pernet = pm_nl_get_pernet_from_msk(msk); + struct mptcp_pm_addr_entry *entry; + bool backup = false; + + rcu_read_lock(); + list_for_each_entry_rcu(entry, &pernet->local_addr_list, list) { + if (mptcp_addresses_equal(&entry->addr, skc, entry->addr.port)) { + backup = !!(entry->flags & MPTCP_PM_ADDR_FLAG_BACKUP); + break; + } + } + rcu_read_unlock(); + + return backup; +} + #define MPTCP_PM_CMD_GRP_OFFSET 0 #define MPTCP_PM_EV_GRP_OFFSET 1 diff --git a/net/mptcp/pm_userspace.c b/net/mptcp/pm_userspace.c index f0a4590506c6..8eaa9fbe3e34 100644 --- a/net/mptcp/pm_userspace.c +++ b/net/mptcp/pm_userspace.c @@ -165,6 +165,24 @@ int mptcp_userspace_pm_get_local_id(struct mptcp_sock *msk, return mptcp_userspace_pm_append_new_local_addr(msk, &new_entry, true); } +bool mptcp_userspace_pm_is_backup(struct mptcp_sock *msk, + struct mptcp_addr_info *skc) +{ + struct mptcp_pm_addr_entry *entry; + bool backup = false; + + spin_lock_bh(&msk->pm.lock); + list_for_each_entry(entry, &msk->pm.userspace_pm_local_addr_list, list) { + if (mptcp_addresses_equal(&entry->addr, skc, false)) { + backup = !!(entry->flags & MPTCP_PM_ADDR_FLAG_BACKUP); + break; + } + } + spin_unlock_bh(&msk->pm.lock); + + return backup; +} + int mptcp_pm_nl_announce_doit(struct sk_buff *skb, struct genl_info *info) { struct nlattr *token = info->attrs[MPTCP_PM_ATTR_TOKEN]; diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h index b8b25124e7de..60c6b073d65f 100644 --- a/net/mptcp/protocol.h +++ b/net/mptcp/protocol.h @@ -1109,6 +1109,9 @@ bool mptcp_pm_rm_addr_signal(struct mptcp_sock *msk, unsigned int remaining, int mptcp_pm_get_local_id(struct mptcp_sock *msk, struct sock_common *skc); int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk, struct mptcp_addr_info *skc); int mptcp_userspace_pm_get_local_id(struct mptcp_sock *msk, struct mptcp_addr_info *skc); +bool mptcp_pm_is_backup(struct mptcp_sock *msk, struct sock_common *skc); +bool mptcp_pm_nl_is_backup(struct mptcp_sock *msk, struct mptcp_addr_info *skc); +bool mptcp_userspace_pm_is_backup(struct mptcp_sock *msk, struct mptcp_addr_info *skc); int mptcp_pm_dump_addr(struct sk_buff *msg, struct netlink_callback *cb); int mptcp_pm_nl_dump_addr(struct sk_buff *msg, struct netlink_callback *cb); diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c index be406197b1c4..0e4b5bfbeaa1 100644 --- a/net/mptcp/subflow.c +++ b/net/mptcp/subflow.c @@ -100,6 +100,7 @@ static struct mptcp_sock *subflow_token_join_request(struct request_sock *req) return NULL; } subflow_req->local_id = local_id; + subflow_req->request_bkup = mptcp_pm_is_backup(msk, (struct sock_common *)req); return msk; } @@ -620,6 +621,8 @@ static int subflow_chk_local_id(struct sock *sk) return err; subflow_set_local_id(subflow, err); + subflow->request_bkup = mptcp_pm_is_backup(msk, (struct sock_common *)sk); + return 0; }

4 months, 2 weeks

2
1
0 0

[tip: x86/urgent] x86/paravirt: Fix incorrect virt spinlock setting on bare metal

by tip-bot2 for Chen Yu

The following commit has been merged into the x86/urgent branch of tip: Commit-ID: e639222a51196c69c70b49b67098ce2f9919ed08 Gitweb: https://git.kernel.org/tip/e639222a51196c69c70b49b67098ce2f9919ed08 Author: Chen Yu <yu.c.chen(a)intel.com> AuthorDate: Tue, 06 Aug 2024 19:22:07 +08:00 Committer: Thomas Gleixner <tglx(a)linutronix.de> CommitterDate: Wed, 07 Aug 2024 20:04:38 +02:00 x86/paravirt: Fix incorrect virt spinlock setting on bare metal The kernel can change spinlock behavior when running as a guest. But this guest-friendly behavior causes performance problems on bare metal. The kernel uses a static key to switch between the two modes. In theory, the static key is enabled by default (run in guest mode) and should be disabled for bare metal (and in some guests that want native behavior or paravirt spinlock). A performance drop is reported when running encode/decode workload and BenchSEE cache sub-workload. Bisect points to commit ce0a1b608bfc ("x86/paravirt: Silence unused native_pv_lock_init() function warning"). When CONFIG_PARAVIRT_SPINLOCKS is disabled the virt_spin_lock_key is incorrectly set to true on bare metal. The qspinlock degenerates to test-and-set spinlock, which decreases the performance on bare metal. Set the default value of virt_spin_lock_key to false. If booting in a VM, enable this key. Later during the VM initialization, if other high-efficient spinlock is preferred (e.g. paravirt-spinlock), or the user wants the native qspinlock (via nopvspin boot commandline), the virt_spin_lock_key is disabled accordingly. This results in the following decision matrix: X86_FEATURE_HYPERVISOR Y Y Y N CONFIG_PARAVIRT_SPINLOCKS Y Y N Y/N PV spinlock Y N N Y/N virt_spin_lock_key N Y/N Y N Fixes: ce0a1b608bfc ("x86/paravirt: Silence unused native_pv_lock_init() function warning") Reported-by: Prem Nath Dey <prem.nath.dey(a)intel.com> Reported-by: Xiaoping Zhou <xiaoping.zhou(a)intel.com> Suggested-by: Dave Hansen <dave.hansen(a)linux.intel.com> Suggested-by: Qiuxu Zhuo <qiuxu.zhuo(a)intel.com> Suggested-by: Nikolay Borisov <nik.borisov(a)suse.com> Signed-off-by: Chen Yu <yu.c.chen(a)intel.com> Signed-off-by: Thomas Gleixner <tglx(a)linutronix.de> Reviewed-by: Nikolay Borisov <nik.borisov(a)suse.com> Cc: stable(a)vger.kernel.org Link: https://lore.kernel.org/all/20240806112207.29792-1-yu.c.chen@intel.com --- arch/x86/include/asm/qspinlock.h | 12 +++++++----- arch/x86/kernel/paravirt.c | 7 +++---- 2 files changed, 10 insertions(+), 9 deletions(-) diff --git a/arch/x86/include/asm/qspinlock.h b/arch/x86/include/asm/qspinlock.h index a053c12..68da67d 100644 --- a/arch/x86/include/asm/qspinlock.h +++ b/arch/x86/include/asm/qspinlock.h @@ -66,13 +66,15 @@ static inline bool vcpu_is_preempted(long cpu) #ifdef CONFIG_PARAVIRT /* - * virt_spin_lock_key - enables (by default) the virt_spin_lock() hijack. + * virt_spin_lock_key - disables by default the virt_spin_lock() hijack. * - * Native (and PV wanting native due to vCPU pinning) should disable this key. - * It is done in this backwards fashion to only have a single direction change, - * which removes ordering between native_pv_spin_init() and HV setup. + * Native (and PV wanting native due to vCPU pinning) should keep this key + * disabled. Native does not touch the key. + * + * When in a guest then native_pv_lock_init() enables the key first and + * KVM/XEN might conditionally disable it later in the boot process again. */ -DECLARE_STATIC_KEY_TRUE(virt_spin_lock_key); +DECLARE_STATIC_KEY_FALSE(virt_spin_lock_key); /* * Shortcut for the queued_spin_lock_slowpath() function that allows diff --git a/arch/x86/kernel/paravirt.c b/arch/x86/kernel/paravirt.c index 5358d43..fec3815 100644 --- a/arch/x86/kernel/paravirt.c +++ b/arch/x86/kernel/paravirt.c @@ -51,13 +51,12 @@ DEFINE_ASM_FUNC(pv_native_irq_enable, "sti", .noinstr.text); DEFINE_ASM_FUNC(pv_native_read_cr2, "mov %cr2, %rax", .noinstr.text); #endif -DEFINE_STATIC_KEY_TRUE(virt_spin_lock_key); +DEFINE_STATIC_KEY_FALSE(virt_spin_lock_key); void __init native_pv_lock_init(void) { - if (IS_ENABLED(CONFIG_PARAVIRT_SPINLOCKS) && - !boot_cpu_has(X86_FEATURE_HYPERVISOR)) - static_branch_disable(&virt_spin_lock_key); + if (boot_cpu_has(X86_FEATURE_HYPERVISOR)) + static_branch_enable(&virt_spin_lock_key); } static void native_tlb_remove_table(struct mmu_gather *tlb, void *table)

4 months, 2 weeks

1
0
0 0

[PATCH v3 bpf-next 01/10] lib/buildid: harden build ID parsing logic

by Andrii Nakryiko

Harden build ID parsing logic, adding explicit READ_ONCE() where it's important to have a consistent value read and validated just once. Fixes tag below points to the code that moved this code into lib/buildid.c, and then subsequently was used in perf subsystem, making this code exposed to perf_event_open() users in v5.12+. Cc: stable(a)vger.kernel.org Cc: Jann Horn <jannh(a)google.com> Suggested-by: Andi Kleen <ak(a)linux.intel.com> Fixes: bd7525dacd7e ("bpf: Move stack_map_get_build_id into lib") Signed-off-by: Andrii Nakryiko <andrii(a)kernel.org> --- lib/buildid.c | 51 +++++++++++++++++++++++++++------------------------ 1 file changed, 27 insertions(+), 24 deletions(-) diff --git a/lib/buildid.c b/lib/buildid.c index e02b5507418b..d21d86f6c19a 100644 --- a/lib/buildid.c +++ b/lib/buildid.c @@ -18,28 +18,29 @@ static int parse_build_id_buf(unsigned char *build_id, const void *note_start, Elf32_Word note_size) { + const char note_name[] = "GNU"; + const size_t note_name_sz = sizeof(note_name); Elf32_Word note_offs = 0, new_offs; + u32 name_sz, desc_sz; + const char *data; while (note_offs + sizeof(Elf32_Nhdr) < note_size) { Elf32_Nhdr *nhdr = (Elf32_Nhdr *)(note_start + note_offs); + name_sz = READ_ONCE(nhdr->n_namesz); + desc_sz = READ_ONCE(nhdr->n_descsz); if (nhdr->n_type == BUILD_ID && - nhdr->n_namesz == sizeof("GNU") && - !strcmp((char *)(nhdr + 1), "GNU") && - nhdr->n_descsz > 0 && - nhdr->n_descsz <= BUILD_ID_SIZE_MAX) { - memcpy(build_id, - note_start + note_offs + - ALIGN(sizeof("GNU"), 4) + sizeof(Elf32_Nhdr), - nhdr->n_descsz); - memset(build_id + nhdr->n_descsz, 0, - BUILD_ID_SIZE_MAX - nhdr->n_descsz); + name_sz == note_name_sz && + strcmp((char *)(nhdr + 1), note_name) == 0 && + desc_sz > 0 && desc_sz <= BUILD_ID_SIZE_MAX) { + data = note_start + note_offs + ALIGN(note_name_sz, 4); + memcpy(build_id, data, desc_sz); + memset(build_id + desc_sz, 0, BUILD_ID_SIZE_MAX - desc_sz); if (size) - *size = nhdr->n_descsz; + *size = desc_sz; return 0; } - new_offs = note_offs + sizeof(Elf32_Nhdr) + - ALIGN(nhdr->n_namesz, 4) + ALIGN(nhdr->n_descsz, 4); + new_offs = note_offs + sizeof(Elf32_Nhdr) + ALIGN(name_sz, 4) + ALIGN(desc_sz, 4); if (new_offs <= note_offs) /* overflow */ break; note_offs = new_offs; @@ -71,7 +72,7 @@ static int get_build_id_32(const void *page_addr, unsigned char *build_id, { Elf32_Ehdr *ehdr = (Elf32_Ehdr *)page_addr; Elf32_Phdr *phdr; - int i; + __u32 i, phnum; /* * FIXME @@ -80,9 +81,10 @@ static int get_build_id_32(const void *page_addr, unsigned char *build_id, */ if (ehdr->e_phoff != sizeof(Elf32_Ehdr)) return -EINVAL; + + phnum = READ_ONCE(ehdr->e_phnum); /* only supports phdr that fits in one page */ - if (ehdr->e_phnum > - (PAGE_SIZE - sizeof(Elf32_Ehdr)) / sizeof(Elf32_Phdr)) + if (phnum > (PAGE_SIZE - sizeof(Elf32_Ehdr)) / sizeof(Elf32_Phdr)) return -EINVAL; phdr = (Elf32_Phdr *)(page_addr + sizeof(Elf32_Ehdr)); @@ -90,8 +92,8 @@ static int get_build_id_32(const void *page_addr, unsigned char *build_id, for (i = 0; i < ehdr->e_phnum; ++i) { if (phdr[i].p_type == PT_NOTE && !parse_build_id(page_addr, build_id, size, - page_addr + phdr[i].p_offset, - phdr[i].p_filesz)) + page_addr + READ_ONCE(phdr[i].p_offset), + READ_ONCE(phdr[i].p_filesz))) return 0; } return -EINVAL; @@ -103,7 +105,7 @@ static int get_build_id_64(const void *page_addr, unsigned char *build_id, { Elf64_Ehdr *ehdr = (Elf64_Ehdr *)page_addr; Elf64_Phdr *phdr; - int i; + __u32 i, phnum; /* * FIXME @@ -112,18 +114,19 @@ static int get_build_id_64(const void *page_addr, unsigned char *build_id, */ if (ehdr->e_phoff != sizeof(Elf64_Ehdr)) return -EINVAL; + + phnum = READ_ONCE(ehdr->e_phnum); /* only supports phdr that fits in one page */ - if (ehdr->e_phnum > - (PAGE_SIZE - sizeof(Elf64_Ehdr)) / sizeof(Elf64_Phdr)) + if (phnum > (PAGE_SIZE - sizeof(Elf64_Ehdr)) / sizeof(Elf64_Phdr)) return -EINVAL; phdr = (Elf64_Phdr *)(page_addr + sizeof(Elf64_Ehdr)); - for (i = 0; i < ehdr->e_phnum; ++i) { + for (i = 0; i < phnum; ++i) { if (phdr[i].p_type == PT_NOTE && !parse_build_id(page_addr, build_id, size, - page_addr + phdr[i].p_offset, - phdr[i].p_filesz)) + page_addr + READ_ONCE(phdr[i].p_offset), + READ_ONCE(phdr[i].p_filesz))) return 0; } return -EINVAL; -- 2.43.0

4 months, 2 weeks

4
4
0 0

FAILED: patch "[PATCH] btrfs: fix corruption after buffer fault in during direct IO" failed to apply to 6.10-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.10-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. To reproduce the conflict and resubmit, you may use the following commands: git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-6.10.y git checkout FETCH_HEAD git cherry-pick -x 939b656bc8ab203fdbde26ccac22bcb7f0985be5 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2024080730-deafness-structure-9630@gregkh' --subject-prefix 'PATCH 6.10.y' HEAD^.. Possible dependencies: 939b656bc8ab ("btrfs: fix corruption after buffer fault in during direct IO append write") 9aa29a20b700 ("btrfs: move the direct IO code into its own file") 04ef7631bfa5 ("btrfs: cleanup duplicated parameters related to btrfs_create_dio_extent()") 9fec848b3a33 ("btrfs: cleanup duplicated parameters related to create_io_em()") e9ea31fb5c1f ("btrfs: cleanup duplicated parameters related to btrfs_alloc_ordered_extent") cdc627e65c7e ("btrfs: cleanup duplicated parameters related to can_nocow_file_extent_args") c77a8c61002e ("btrfs: remove extent_map::block_start member") e28b851ed9b2 ("btrfs: remove extent_map::block_len member") 4aa7b5d1784f ("btrfs: remove extent_map::orig_start member") 3f255ece2f1e ("btrfs: introduce extra sanity checks for extent maps") 3d2ac9922465 ("btrfs: introduce new members for extent_map") 87a6962f73b1 ("btrfs: export the expected file extent through can_nocow_extent()") e8fe524da027 ("btrfs: rename extent_map::orig_block_len to disk_num_bytes") 8996f61ab9ff ("btrfs: move fiemap code into its own file") 56b7169f691c ("btrfs: use a btrfs_inode local variable at btrfs_sync_file()") e641e323abb3 ("btrfs: pass a btrfs_inode to btrfs_wait_ordered_range()") cef2daba4268 ("btrfs: pass a btrfs_inode to btrfs_fdatawrite_range()") 4e660ca3a98d ("btrfs: use a regular rb_root instead of cached rb_root for extent_map_tree") 7f5830bc964d ("btrfs: rename rb_root member of extent_map_tree from map to root") thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 939b656bc8ab203fdbde26ccac22bcb7f0985be5 Mon Sep 17 00:00:00 2001 From: Filipe Manana <fdmanana(a)suse.com> Date: Fri, 26 Jul 2024 11:12:52 +0100 Subject: [PATCH] btrfs: fix corruption after buffer fault in during direct IO append write During an append (O_APPEND write flag) direct IO write if the input buffer was not previously faulted in, we can corrupt the file in a way that the final size is unexpected and it includes an unexpected hole. The problem happens like this: 1) We have an empty file, with size 0, for example; 2) We do an O_APPEND direct IO with a length of 4096 bytes and the input buffer is not currently faulted in; 3) We enter btrfs_direct_write(), lock the inode and call generic_write_checks(), which calls generic_write_checks_count(), and that function sets the iocb position to 0 with the following code: if (iocb->ki_flags & IOCB_APPEND) iocb->ki_pos = i_size_read(inode); 4) We call btrfs_dio_write() and enter into iomap, which will end up calling btrfs_dio_iomap_begin() and that calls btrfs_get_blocks_direct_write(), where we update the i_size of the inode to 4096 bytes; 5) After btrfs_dio_iomap_begin() returns, iomap will attempt to access the page of the write input buffer (at iomap_dio_bio_iter(), with a call to bio_iov_iter_get_pages()) and fail with -EFAULT, which gets returned to btrfs at btrfs_direct_write() via btrfs_dio_write(); 6) At btrfs_direct_write() we get the -EFAULT error, unlock the inode, fault in the write buffer and then goto to the label 'relock'; 7) We lock again the inode, do all the necessary checks again and call again generic_write_checks(), which calls generic_write_checks_count() again, and there we set the iocb's position to 4K, which is the current i_size of the inode, with the following code pointed above: if (iocb->ki_flags & IOCB_APPEND) iocb->ki_pos = i_size_read(inode); 8) Then we go again to btrfs_dio_write() and enter iomap and the write succeeds, but it wrote to the file range [4K, 8K), leaving a hole in the [0, 4K) range and an i_size of 8K, which goes against the expectations of having the data written to the range [0, 4K) and get an i_size of 4K. Fix this by not unlocking the inode before faulting in the input buffer, in case we get -EFAULT or an incomplete write, and not jumping to the 'relock' label after faulting in the buffer - instead jump to a location immediately before calling iomap, skipping all the write checks and relocking. This solves this problem and it's fine even in case the input buffer is memory mapped to the same file range, since only holding the range locked in the inode's io tree can cause a deadlock, it's safe to keep the inode lock (VFS lock), as was fixed and described in commit 51bd9563b678 ("btrfs: fix deadlock due to page faults during direct IO reads and writes"). A sample reproducer provided by a reporter is the following: $ cat test.c #ifndef _GNU_SOURCE #define _GNU_SOURCE #endif #include <fcntl.h> #include <stdio.h> #include <sys/mman.h> #include <sys/stat.h> #include <unistd.h> int main(int argc, char *argv[]) { if (argc < 2) { fprintf(stderr, "Usage: %s <test file>\n", argv[0]); return 1; } int fd = open(argv[1], O_WRONLY | O_CREAT | O_TRUNC | O_DIRECT | O_APPEND, 0644); if (fd < 0) { perror("creating test file"); return 1; } char *buf = mmap(NULL, 4096, PROT_READ, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); ssize_t ret = write(fd, buf, 4096); if (ret < 0) { perror("pwritev2"); return 1; } struct stat stbuf; ret = fstat(fd, &stbuf); if (ret < 0) { perror("stat"); return 1; } printf("size: %llu\n", (unsigned long long)stbuf.st_size); return stbuf.st_size == 4096 ? 0 : 1; } A test case for fstests will be sent soon. Reported-by: Hanna Czenczek <hreitz(a)redhat.com> Link: https://lore.kernel.org/linux-btrfs/0b841d46-12fe-4e64-9abb-871d8d0de271@re… Fixes: 8184620ae212 ("btrfs: fix lost file sync on direct IO write with nowait and dsync iocb") CC: stable(a)vger.kernel.org # 6.1+ Tested-by: Hanna Czenczek <hreitz(a)redhat.com> Reviewed-by: Josef Bacik <josef(a)toxicpanda.com> Signed-off-by: Filipe Manana <fdmanana(a)suse.com> Signed-off-by: David Sterba <dsterba(a)suse.com> diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index c8568b1a61c4..75fa563e4cac 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -459,6 +459,7 @@ struct btrfs_file_private { void *filldir_buf; u64 last_index; struct extent_state *llseek_cached_state; + bool fsync_skip_inode_lock; }; static inline u32 BTRFS_LEAF_DATA_SIZE(const struct btrfs_fs_info *info) diff --git a/fs/btrfs/direct-io.c b/fs/btrfs/direct-io.c index f9fb2db6a1e4..67adbe9d294a 100644 --- a/fs/btrfs/direct-io.c +++ b/fs/btrfs/direct-io.c @@ -856,21 +856,37 @@ ssize_t btrfs_direct_write(struct kiocb *iocb, struct iov_iter *from) * So here we disable page faults in the iov_iter and then retry if we * got -EFAULT, faulting in the pages before the retry. */ +again: from->nofault = true; dio = btrfs_dio_write(iocb, from, written); from->nofault = false; - /* - * iomap_dio_complete() will call btrfs_sync_file() if we have a dsync - * iocb, and that needs to lock the inode. So unlock it before calling - * iomap_dio_complete() to avoid a deadlock. - */ - btrfs_inode_unlock(BTRFS_I(inode), ilock_flags); - - if (IS_ERR_OR_NULL(dio)) + if (IS_ERR_OR_NULL(dio)) { ret = PTR_ERR_OR_ZERO(dio); - else + } else { + struct btrfs_file_private stack_private = { 0 }; + struct btrfs_file_private *private; + const bool have_private = (file->private_data != NULL); + + if (!have_private) + file->private_data = &stack_private; + + /* + * If we have a synchronous write, we must make sure the fsync + * triggered by the iomap_dio_complete() call below doesn't + * deadlock on the inode lock - we are already holding it and we + * can't call it after unlocking because we may need to complete + * partial writes due to the input buffer (or parts of it) not + * being already faulted in. + */ + private = file->private_data; + private->fsync_skip_inode_lock = true; ret = iomap_dio_complete(dio); + private->fsync_skip_inode_lock = false; + + if (!have_private) + file->private_data = NULL; + } /* No increment (+=) because iomap returns a cumulative value. */ if (ret > 0) @@ -897,10 +913,12 @@ ssize_t btrfs_direct_write(struct kiocb *iocb, struct iov_iter *from) } else { fault_in_iov_iter_readable(from, left); prev_left = left; - goto relock; + goto again; } } + btrfs_inode_unlock(BTRFS_I(inode), ilock_flags); + /* * If 'ret' is -ENOTBLK or we have not written all data, then it means * we must fallback to buffered IO. diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 21381de906f6..9f10a9f23fcc 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -1603,6 +1603,7 @@ static inline bool skip_inode_logging(const struct btrfs_log_ctx *ctx) */ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) { + struct btrfs_file_private *private = file->private_data; struct dentry *dentry = file_dentry(file); struct btrfs_inode *inode = BTRFS_I(d_inode(dentry)); struct btrfs_root *root = inode->root; @@ -1612,6 +1613,7 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) int ret = 0, err; u64 len; bool full_sync; + const bool skip_ilock = (private ? private->fsync_skip_inode_lock : false); trace_btrfs_sync_file(file, datasync); @@ -1639,7 +1641,10 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) if (ret) goto out; - btrfs_inode_lock(inode, BTRFS_ILOCK_MMAP); + if (skip_ilock) + down_write(&inode->i_mmap_lock); + else + btrfs_inode_lock(inode, BTRFS_ILOCK_MMAP); atomic_inc(&root->log_batch); @@ -1663,7 +1668,10 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) */ ret = start_ordered_ops(inode, start, end); if (ret) { - btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP); + if (skip_ilock) + up_write(&inode->i_mmap_lock); + else + btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP); goto out; } @@ -1788,7 +1796,10 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) * file again, but that will end up using the synchronization * inside btrfs_sync_log to keep things safe. */ - btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP); + if (skip_ilock) + up_write(&inode->i_mmap_lock); + else + btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP); if (ret == BTRFS_NO_LOG_SYNC) { ret = btrfs_end_transaction(trans);

4 months, 2 weeks

3
3
0 0

Re: [PATCH 6.10 000/123] 6.10.4-rc1 review

by Ronald Warsow

Hi Greg no regressions here on x86_64 (RKL, Intel 11th Gen. CPU) Thanks Tested-by: Ronald Warsow <rwarsow(a)gmx.de>

4 months, 2 weeks

1
0
0 0

[PATCH for 6.1 stable] btrfs: fix corruption after buffer fault in during direct IO append write

by fdmanana＠kernel.org

From: Filipe Manana <fdmanana(a)suse.com> commit 939b656bc8ab203fdbde26ccac22bcb7f0985be5 upstream. During an append (O_APPEND write flag) direct IO write if the input buffer was not previously faulted in, we can corrupt the file in a way that the final size is unexpected and it includes an unexpected hole. The problem happens like this: 1) We have an empty file, with size 0, for example; 2) We do an O_APPEND direct IO with a length of 4096 bytes and the input buffer is not currently faulted in; 3) We enter btrfs_direct_write(), lock the inode and call generic_write_checks(), which calls generic_write_checks_count(), and that function sets the iocb position to 0 with the following code: if (iocb->ki_flags & IOCB_APPEND) iocb->ki_pos = i_size_read(inode); 4) We call btrfs_dio_write() and enter into iomap, which will end up calling btrfs_dio_iomap_begin() and that calls btrfs_get_blocks_direct_write(), where we update the i_size of the inode to 4096 bytes; 5) After btrfs_dio_iomap_begin() returns, iomap will attempt to access the page of the write input buffer (at iomap_dio_bio_iter(), with a call to bio_iov_iter_get_pages()) and fail with -EFAULT, which gets returned to btrfs at btrfs_direct_write() via btrfs_dio_write(); 6) At btrfs_direct_write() we get the -EFAULT error, unlock the inode, fault in the write buffer and then goto to the label 'relock'; 7) We lock again the inode, do all the necessary checks again and call again generic_write_checks(), which calls generic_write_checks_count() again, and there we set the iocb's position to 4K, which is the current i_size of the inode, with the following code pointed above: if (iocb->ki_flags & IOCB_APPEND) iocb->ki_pos = i_size_read(inode); 8) Then we go again to btrfs_dio_write() and enter iomap and the write succeeds, but it wrote to the file range [4K, 8K[, leaving a hole in the [0, 4K[ range and an i_size of 8K, which goes against the expections of having the data written to the range [0, 4K[ and get an i_size of 4K. Fix this by not unlocking the inode before faulting in the input buffer, in case we get -EFAULT or an incomplete write, and not jumping to the 'relock' label after faulting in the buffer - instead jump to a location immediately before calling iomap, skipping all the write checks and relocking. This solves this problem and it's fine even in case the input buffer is memory mapped to the same file range, since only holding the range locked in the inode's io tree can cause a deadlock, it's safe to keep the inode lock (VFS lock), as was fixed and described in commit 51bd9563b678 ("btrfs: fix deadlock due to page faults during direct IO reads and writes"). A sample reproducer provided by a reporter is the following: $ cat test.c #ifndef _GNU_SOURCE #define _GNU_SOURCE #endif #include <fcntl.h> #include <stdio.h> #include <sys/mman.h> #include <sys/stat.h> #include <unistd.h> int main(int argc, char *argv[]) { if (argc < 2) { fprintf(stderr, "Usage: %s <test file>\n", argv[0]); return 1; } int fd = open(argv[1], O_WRONLY | O_CREAT | O_TRUNC | O_DIRECT | O_APPEND, 0644); if (fd < 0) { perror("creating test file"); return 1; } char *buf = mmap(NULL, 4096, PROT_READ, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); ssize_t ret = write(fd, buf, 4096); if (ret < 0) { perror("pwritev2"); return 1; } struct stat stbuf; ret = fstat(fd, &stbuf); if (ret < 0) { perror("stat"); return 1; } printf("size: %llu\n", (unsigned long long)stbuf.st_size); return stbuf.st_size == 4096 ? 0 : 1; } A test case for fstests will be sent soon. Reported-by: Hanna Czenczek <hreitz(a)redhat.com> Link: https://lore.kernel.org/linux-btrfs/0b841d46-12fe-4e64-9abb-871d8d0de271@re… Fixes: 8184620ae212 ("btrfs: fix lost file sync on direct IO write with nowait and dsync iocb") Signed-off-by: Filipe Manana <fdmanana(a)suse.com> --- fs/btrfs/ctree.h | 1 + fs/btrfs/file.c | 55 ++++++++++++++++++++++++++++++++++++------------ 2 files changed, 43 insertions(+), 13 deletions(-) diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index cca1acf2e037..853b1f96b1fd 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1553,6 +1553,7 @@ struct btrfs_drop_extents_args { struct btrfs_file_private { void *filldir_buf; u64 last_index; + bool fsync_skip_inode_lock; }; diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 1783a0fbf166..7c3ae295fdb5 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -1526,21 +1526,37 @@ static ssize_t btrfs_direct_write(struct kiocb *iocb, struct iov_iter *from) * So here we disable page faults in the iov_iter and then retry if we * got -EFAULT, faulting in the pages before the retry. */ +again: from->nofault = true; dio = btrfs_dio_write(iocb, from, written); from->nofault = false; - /* - * iomap_dio_complete() will call btrfs_sync_file() if we have a dsync - * iocb, and that needs to lock the inode. So unlock it before calling - * iomap_dio_complete() to avoid a deadlock. - */ - btrfs_inode_unlock(inode, ilock_flags); - - if (IS_ERR_OR_NULL(dio)) + if (IS_ERR_OR_NULL(dio)) { err = PTR_ERR_OR_ZERO(dio); - else + } else { + struct btrfs_file_private stack_private = { 0 }; + struct btrfs_file_private *private; + const bool have_private = (file->private_data != NULL); + + if (!have_private) + file->private_data = &stack_private; + + /* + * If we have a synchoronous write, we must make sure the fsync + * triggered by the iomap_dio_complete() call below doesn't + * deadlock on the inode lock - we are already holding it and we + * can't call it after unlocking because we may need to complete + * partial writes due to the input buffer (or parts of it) not + * being already faulted in. + */ + private = file->private_data; + private->fsync_skip_inode_lock = true; err = iomap_dio_complete(dio); + private->fsync_skip_inode_lock = false; + + if (!have_private) + file->private_data = NULL; + } /* No increment (+=) because iomap returns a cumulative value. */ if (err > 0) @@ -1567,10 +1583,12 @@ static ssize_t btrfs_direct_write(struct kiocb *iocb, struct iov_iter *from) } else { fault_in_iov_iter_readable(from, left); prev_left = left; - goto relock; + goto again; } } + btrfs_inode_unlock(inode, ilock_flags); + /* * If 'err' is -ENOTBLK or we have not written all data, then it means * we must fallback to buffered IO. @@ -1777,6 +1795,7 @@ static inline bool skip_inode_logging(const struct btrfs_log_ctx *ctx) */ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) { + struct btrfs_file_private *private = file->private_data; struct dentry *dentry = file_dentry(file); struct inode *inode = d_inode(dentry); struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); @@ -1786,6 +1805,7 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) int ret = 0, err; u64 len; bool full_sync; + const bool skip_ilock = (private ? private->fsync_skip_inode_lock : false); trace_btrfs_sync_file(file, datasync); @@ -1813,7 +1833,10 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) if (ret) goto out; - btrfs_inode_lock(inode, BTRFS_ILOCK_MMAP); + if (skip_ilock) + down_write(&BTRFS_I(inode)->i_mmap_lock); + else + btrfs_inode_lock(inode, BTRFS_ILOCK_MMAP); atomic_inc(&root->log_batch); @@ -1837,7 +1860,10 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) */ ret = start_ordered_ops(inode, start, end); if (ret) { - btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP); + if (skip_ilock) + up_write(&BTRFS_I(inode)->i_mmap_lock); + else + btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP); goto out; } @@ -1940,7 +1966,10 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) * file again, but that will end up using the synchronization * inside btrfs_sync_log to keep things safe. */ - btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP); + if (skip_ilock) + up_write(&BTRFS_I(inode)->i_mmap_lock); + else + btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP); if (ret == BTRFS_NO_LOG_SYNC) { ret = btrfs_end_transaction(trans); -- 2.43.0

4 months, 2 weeks

1
0
0 0

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror August 2024