Skip to contentSkip to conversation

for-next: KCSAN failures on 6130a25681d4 (kdave/for-next) Merge branch 'for-next-next-v5.20-20220804' into for-next-20220804

DISCUSSION Zygo Blaxell 4 participants 5 messages Aug 9, 2022 – Sep 5, 2022 Btrfs file system

lore
Share & tools
lore
#0 lore raw

Some KCSAN complaints I found while testing for other things...

Here's one related to extent refs:

	[ 2799.504795][ T5562] ==================================================================
	[ 2799.507470][ T5562] BUG: KCSAN: data-race in btrfs_block_rsv_release / btrfs_update_delayed_refs_rsv
	[ 2799.509416][ T5562] 
	[ 2799.509699][ T5562] write to 0xffff9c80059504c8 of 1 bytes by task 104535 on cpu 1:
	[ 2799.510634][ T5562]  btrfs_update_delayed_refs_rsv+0xcf/0x110
	[ 2799.511422][ T5562]  btrfs_add_delayed_tree_ref+0x393/0x690
	[ 2799.513587][ T5562]  btrfs_inc_extent_ref+0x8e/0xa0
	[ 2799.515080][ T5562]  __btrfs_mod_ref+0x3ca/0x6f0
	[ 2799.516543][ T5562]  btrfs_inc_ref+0x3d/0x70
	[ 2799.517010][ T5562]  update_ref_for_cow+0x41a/0x530
	[ 2799.519588][ T5562]  __btrfs_cow_block+0x4a6/0xae0
	[ 2799.521443][ T5562]  btrfs_cow_block+0x1da/0x3f0
	[ 2799.523088][ T5562]  btrfs_search_slot+0xb52/0x1460
	[ 2799.524794][ T5562]  btrfs_lookup_inode+0x5f/0x190
	[ 2799.526496][ T5562]  __btrfs_update_delayed_inode+0xed/0x4b0
	[ 2799.528514][ T5562]  btrfs_commit_inode_delayed_inode+0x1d8/0x1e0
	[ 2799.530614][ T5562]  btrfs_evict_inode+0x5dc/0x800
	[ 2799.532280][ T5562]  evict+0x1ad/0x330
	[ 2799.533807][ T5562]  iput.part.0+0x2e3/0x410
	[ 2799.535296][ T5562]  iput+0x3c/0x50
	[ 2799.535771][ T5562]  dentry_unlink_inode+0x1f4/0x280
	[ 2799.536427][ T5562]  __dentry_kill+0x21f/0x330
	[ 2799.538323][ T5562]  dput+0x44e/0x810
	[ 2799.538810][ T5562]  do_renameat2+0x4d5/0x900
	[ 2799.539374][ T5562]  __x64_sys_rename+0x67/0x90
	[ 2799.541644][ T5562]  do_syscall_64+0x5f/0xd0
	[ 2799.543103][ T5562]  entry_SYSCALL_64_after_hwframe+0x63/0xcd
	[ 2799.544831][ T5562] 
	[ 2799.545138][ T5562] INFO: lockdep is turned off.
	[ 2799.545732][ T5562] irq event stamp: 0
	[ 2799.546233][ T5562] hardirqs last  enabled at (0): [<0000000000000000>] 0x0
	[ 2799.547133][ T5562] hardirqs last disabled at (0): [<ffffffffa1127cca>] copy_process+0xe6a/0x3760
	[ 2799.548295][ T5562] softirqs last  enabled at (0): [<ffffffffa1127cca>] copy_process+0xe6a/0x3760
	[ 2799.549457][ T5562] softirqs last disabled at (0): [<0000000000000000>] 0x0
	[ 2799.550353][ T5562] 
	[ 2799.550654][ T5562] read to 0xffff9c80059504c8 of 1 bytes by task 5562 on cpu 3:
	[ 2799.551641][ T5562]  btrfs_block_rsv_release+0x5b/0x3f0
	[ 2799.552369][ T5562]  btrfs_inode_rsv_release+0x78/0x120
	[ 2799.553100][ T5562]  btrfs_delalloc_release_metadata+0xc1/0xf0
	[ 2799.553901][ T5562]  btrfs_clear_delalloc_extent+0x1b3/0x4b0
	[ 2799.554684][ T5562]  clear_state_bit+0x273/0x2c0
	[ 2799.555335][ T5562]  __clear_extent_bit+0x2f7/0x630
	[ 2799.556028][ T5562]  clear_extent_bit+0x54/0x80
	[ 2799.556664][ T5562]  btrfs_dirty_pages+0xe6/0x1d0
	[ 2799.557438][ T5562]  btrfs_buffered_write+0x6d1/0xa90
	[ 2799.558333][ T5562]  btrfs_do_write_iter+0x464/0x8a0
	[ 2799.559223][ T5562]  btrfs_file_write_iter+0x29/0x50
	[ 2799.560184][ T5562]  new_sync_write+0x204/0x300
	[ 2799.561071][ T5562]  vfs_write+0x457/0x5c0
	[ 2799.561876][ T5562]  ksys_write+0xb2/0x150
	[ 2799.562622][ T5562]  __x64_sys_write+0x4b/0x70
	[ 2799.563516][ T5562]  do_syscall_64+0x5f/0xd0
	[ 2799.564315][ T5562]  entry_SYSCALL_64_after_hwframe+0x63/0xcd
	[ 2799.565431][ T5562] 
	[ 2799.565826][ T5562] INFO: lockdep is turned off.
	[ 2799.566683][ T5562] irq event stamp: 1566656
	[ 2799.567447][ T5562] hardirqs last  enabled at (1566655): [<ffffffffa30ce51d>] _raw_spin_unlock_irqrestore+0x7d/0xa0
	[ 2799.569315][ T5562] hardirqs last disabled at (1566656): [<ffffffffa30bee38>] __schedule+0x9a8/0x18c0
	[ 2799.570991][ T5562] softirqs last  enabled at (1565224): [<ffffffffa113ad8f>] __irq_exit_rcu+0xef/0x160
	[ 2799.572340][ T5562] softirqs last disabled at (1565219): [<ffffffffa113ad8f>] __irq_exit_rcu+0xef/0x160
	[ 2799.573754][ T5562] 
	[ 2799.574131][ T5562] Reported by Kernel Concurrency Sanitizer on:
	[ 2799.574947][ T5562] CPU: 3 PID: 5562 Comm: tee Tainted: G      D           5.19.0-466d9d7ea677-for-next+ #85 89955463945a81b56a449b1f12383cf0d5e6b898
	[ 2799.576954][ T5562] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-2 04/01/2014
	[ 2799.578369][ T5562] ==================================================================

Writer is:

	btrfs_update_delayed_refs_rsv+0xce/0x110:

	btrfs_update_delayed_refs_rsv at fs/btrfs/delayed-ref.c:135
	 130            if (btrfs_test_opt(fs_info, FREE_SPACE_TREE))
	 131                    num_bytes *= 2;
	 132    
	 133            spin_lock(&delayed_rsv->lock);
	 134            delayed_rsv->size += num_bytes;
	>135<           delayed_rsv->full = false;
	 136            spin_unlock(&delayed_rsv->lock);
	 137            trans->delayed_ref_updates = 0;
	 138    }
	 139    
	 140    /**

Reader is:

	btrfs_block_rsv_release+0x5b/0x3f0:

	btrfs_block_rsv_release at fs/btrfs/block-rsv.c:289 (discriminator 1)
	 284             * If we are the delayed_rsv then push to the global rsv, otherwise dump
	 285             * into the delayed rsv if it is not full.
	 286             */
	 287            if (block_rsv == delayed_rsv)
	 288                    target = global_rsv;
	>289<           else if (block_rsv != global_rsv && !delayed_rsv->full)
	 290                    target = delayed_rsv;
	 291    
	 292            if (target && block_rsv->space_info != target->space_info)
	 293                    target = NULL;
	 294    

Another, this time with btrfs_block_rsv_release racing with itself:

	[ 2785.914537][ T3379] ==================================================================
	[ 2785.916300][ T3379] BUG: KCSAN: data-race in btrfs_block_rsv_release / btrfs_block_rsv_release
	[ 2785.918215][ T3379] 
	[ 2785.918641][ T3379] read to 0xffff9c80059504c8 of 1 bytes by task 6169 on cpu 2:
	[ 2785.920426][ T3379]  btrfs_block_rsv_release+0x5b/0x3f0
	[ 2785.921671][ T3379]  btrfs_trans_release_metadata+0xae/0xe0
	[ 2785.956873][ T3379]  __btrfs_end_transaction+0x80/0x4f0
	[ 2785.958523][ T3379]  btrfs_end_transaction+0x21/0x40
	[ 2785.959223][ T3379]  btrfs_evict_inode+0x7e9/0x800
	[ 2785.959896][ T3379]  evict+0x1ad/0x330
	[ 2785.960430][ T3379]  iput.part.0+0x2e3/0x410
	[ 2785.964122][ T3379]  iput+0x3c/0x50
	[ 2785.967801][ T3379]  dentry_unlink_inode+0x1f4/0x280
	[ 2785.973195][ T3379]  __dentry_kill+0x21f/0x330
	[ 2785.978567][ T3379]  dput+0x44e/0x810
	[ 2785.983573][ T3379]  do_renameat2+0x4d5/0x900
	[ 2786.037153][ T3379]  __x64_sys_rename+0x67/0x90
	[ 2786.037717][ T3379]  do_syscall_64+0x5f/0xd0
	[ 2786.038242][ T3379]  entry_SYSCALL_64_after_hwframe+0x63/0xcd
	[ 2786.038931][ T3379] 
	[ 2786.039208][ T3379] INFO: lockdep is turned off.
	[ 2786.039755][ T3379] irq event stamp: 3250759
	[ 2786.040333][ T3379] hardirqs last  enabled at (3250758): [<ffffffffa30ce452>] _raw_spin_unlock_irq+0x32/0x80
	[ 2786.041701][ T3379] hardirqs last disabled at (3250759): [<ffffffffa30bee38>] __schedule+0x9a8/0x18c0
	[ 2786.043073][ T3379] softirqs last  enabled at (3250361): [<ffffffffa113ad8f>] __irq_exit_rcu+0xef/0x160
	[ 2786.044221][ T3379] softirqs last disabled at (3250352): [<ffffffffa113ad8f>] __irq_exit_rcu+0xef/0x160
	[ 2786.045392][ T3379] 
	[ 2786.045712][ T3379] write to 0xffff9c80059504c8 of 1 bytes by task 3379 on cpu 0:
	[ 2786.046658][ T3379]  btrfs_block_rsv_release+0x23e/0x3f0
	[ 2786.047397][ T3379]  btrfs_delayed_refs_rsv_release+0x75/0xd0
	[ 2786.057504][ T3379]  btrfs_cleanup_ref_head_accounting+0x4f/0x140
	[ 2786.058314][ T3379]  __btrfs_run_delayed_refs+0x18c9/0x2080
	[ 2786.059279][ T3379]  btrfs_run_delayed_refs+0xcd/0x310
	[ 2786.061723][ T3379]  btrfs_commit_transaction+0x136/0x1800
	[ 2786.064101][ T3379]  transaction_kthread+0x24c/0x2d0
	[ 2786.066161][ T3379]  kthread+0x1ab/0x1e0
	[ 2786.067969][ T3379]  ret_from_fork+0x22/0x30
	[ 2786.070012][ T3379] 
	[ 2786.070391][ T3379] INFO: lockdep is turned off.
	[ 2786.073032][ T3379] irq event stamp: 5460863
	[ 2786.075562][ T3379] hardirqs last  enabled at (5460862): [<ffffffffa30ce51d>] _raw_spin_unlock_irqrestore+0x7d/0xa0
	[ 2786.079952][ T3379] hardirqs last disabled at (5460863): [<ffffffffa30bee38>] __schedule+0x9a8/0x18c0
	[ 2786.084111][ T3379] softirqs last  enabled at (5460221): [<ffffffffa113ad8f>] __irq_exit_rcu+0xef/0x160
	[ 2786.088033][ T3379] softirqs last disabled at (5460214): [<ffffffffa113ad8f>] __irq_exit_rcu+0xef/0x160
	[ 2786.091680][ T3379] 
	[ 2786.092464][ T3379] Reported by Kernel Concurrency Sanitizer on:
	[ 2786.095225][ T3379] CPU: 0 PID: 3379 Comm: btrfs-transacti Tainted: G      D           5.19.0-466d9d7ea677-for-next+ #85 89955463945a81b56a449b1f12383cf0d5e6b898
	[ 2786.100076][ T3379] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-2 04/01/2014
	[ 2786.102614][ T3379] ==================================================================
	btrfs_block_rsv_release+0x23e/0x3f0:

	btrfs_block_rsv_release at fs/btrfs/block-rsv.c:121
	 116            }
	 117            block_rsv->size -= num_bytes;
	 118            if (block_rsv->reserved >= block_rsv->size) {
	 119                    num_bytes = block_rsv->reserved - block_rsv->size;
	 120                    block_rsv->reserved = block_rsv->size;
	>121<                   block_rsv->full = true;
	 122            } else {
	 123                    num_bytes = 0;
	 124            }
	 125            if (block_rsv->qgroup_rsv_reserved >= block_rsv->qgroup_rsv_size) {
	 126                    qgroup_to_release = block_rsv->qgroup_rsv_reserved -

There's another one in start_transaction. It's basically everywhere, and always looking at some struct's ->full member:

	start_transaction+0x8ce/0xa40:

	start_transaction at fs/btrfs/transaction.c:638 (discriminator 1)
	 633                     * worth of delayed refs updates in this trans handle, and
	 634                     * refill that amount for whatever is missing in the reserve.
	 635                     */
	 636                    num_bytes = btrfs_calc_insert_metadata_size(fs_info, num_items);
	 637                    if (flush == BTRFS_RESERVE_FLUSH_ALL &&
	>638<                       delayed_refs_rsv->full == 0) {
	 639                            delayed_refs_bytes = num_bytes;
	 640                            num_bytes <<= 1;
	 641                    }
	 642    
	 643                    /*
#1 lore raw
Quoted 4 lines · Zygo Blaxell

Some KCSAN complaints I found while testing for other things...

Here's one related to extent refs:

It's about the block reserves, nothing to do with extents refs.

These get reported every now and then like here:

https://lore.kernel.org/linux-btrfs/CAAwBoOJDjei5Hnem155N_cJwiEkVwJYvgN-tQrwWbZQGhFU=cA@mail.gmail.com/

It's actually harmless, but if we keep it like this, we'll keep getting reports in the future.

Quoted 186 lines
        [ 2799.504795][ T5562] ==================================================================
        [ 2799.507470][ T5562] BUG: KCSAN: data-race in btrfs_block_rsv_release / btrfs_update_delayed_refs_rsv
        [ 2799.509416][ T5562]
        [ 2799.509699][ T5562] write to 0xffff9c80059504c8 of 1 bytes by task 104535 on cpu 1:
        [ 2799.510634][ T5562]  btrfs_update_delayed_refs_rsv+0xcf/0x110
        [ 2799.511422][ T5562]  btrfs_add_delayed_tree_ref+0x393/0x690
        [ 2799.513587][ T5562]  btrfs_inc_extent_ref+0x8e/0xa0
        [ 2799.515080][ T5562]  __btrfs_mod_ref+0x3ca/0x6f0
        [ 2799.516543][ T5562]  btrfs_inc_ref+0x3d/0x70
        [ 2799.517010][ T5562]  update_ref_for_cow+0x41a/0x530
        [ 2799.519588][ T5562]  __btrfs_cow_block+0x4a6/0xae0
        [ 2799.521443][ T5562]  btrfs_cow_block+0x1da/0x3f0
        [ 2799.523088][ T5562]  btrfs_search_slot+0xb52/0x1460
        [ 2799.524794][ T5562]  btrfs_lookup_inode+0x5f/0x190
        [ 2799.526496][ T5562]  __btrfs_update_delayed_inode+0xed/0x4b0
        [ 2799.528514][ T5562]  btrfs_commit_inode_delayed_inode+0x1d8/0x1e0
        [ 2799.530614][ T5562]  btrfs_evict_inode+0x5dc/0x800
        [ 2799.532280][ T5562]  evict+0x1ad/0x330
        [ 2799.533807][ T5562]  iput.part.0+0x2e3/0x410
        [ 2799.535296][ T5562]  iput+0x3c/0x50
        [ 2799.535771][ T5562]  dentry_unlink_inode+0x1f4/0x280
        [ 2799.536427][ T5562]  __dentry_kill+0x21f/0x330
        [ 2799.538323][ T5562]  dput+0x44e/0x810
        [ 2799.538810][ T5562]  do_renameat2+0x4d5/0x900
        [ 2799.539374][ T5562]  __x64_sys_rename+0x67/0x90
        [ 2799.541644][ T5562]  do_syscall_64+0x5f/0xd0
        [ 2799.543103][ T5562]  entry_SYSCALL_64_after_hwframe+0x63/0xcd
        [ 2799.544831][ T5562]
        [ 2799.545138][ T5562] INFO: lockdep is turned off.
        [ 2799.545732][ T5562] irq event stamp: 0
        [ 2799.546233][ T5562] hardirqs last  enabled at (0): [<0000000000000000>] 0x0
        [ 2799.547133][ T5562] hardirqs last disabled at (0): [<ffffffffa1127cca>] copy_process+0xe6a/0x3760
        [ 2799.548295][ T5562] softirqs last  enabled at (0): [<ffffffffa1127cca>] copy_process+0xe6a/0x3760
        [ 2799.549457][ T5562] softirqs last disabled at (0): [<0000000000000000>] 0x0
        [ 2799.550353][ T5562]
        [ 2799.550654][ T5562] read to 0xffff9c80059504c8 of 1 bytes by task 5562 on cpu 3:
        [ 2799.551641][ T5562]  btrfs_block_rsv_release+0x5b/0x3f0
        [ 2799.552369][ T5562]  btrfs_inode_rsv_release+0x78/0x120
        [ 2799.553100][ T5562]  btrfs_delalloc_release_metadata+0xc1/0xf0
        [ 2799.553901][ T5562]  btrfs_clear_delalloc_extent+0x1b3/0x4b0
        [ 2799.554684][ T5562]  clear_state_bit+0x273/0x2c0
        [ 2799.555335][ T5562]  __clear_extent_bit+0x2f7/0x630
        [ 2799.556028][ T5562]  clear_extent_bit+0x54/0x80
        [ 2799.556664][ T5562]  btrfs_dirty_pages+0xe6/0x1d0
        [ 2799.557438][ T5562]  btrfs_buffered_write+0x6d1/0xa90
        [ 2799.558333][ T5562]  btrfs_do_write_iter+0x464/0x8a0
        [ 2799.559223][ T5562]  btrfs_file_write_iter+0x29/0x50
        [ 2799.560184][ T5562]  new_sync_write+0x204/0x300
        [ 2799.561071][ T5562]  vfs_write+0x457/0x5c0
        [ 2799.561876][ T5562]  ksys_write+0xb2/0x150
        [ 2799.562622][ T5562]  __x64_sys_write+0x4b/0x70
        [ 2799.563516][ T5562]  do_syscall_64+0x5f/0xd0
        [ 2799.564315][ T5562]  entry_SYSCALL_64_after_hwframe+0x63/0xcd
        [ 2799.565431][ T5562]
        [ 2799.565826][ T5562] INFO: lockdep is turned off.
        [ 2799.566683][ T5562] irq event stamp: 1566656
        [ 2799.567447][ T5562] hardirqs last  enabled at (1566655): [<ffffffffa30ce51d>] _raw_spin_unlock_irqrestore+0x7d/0xa0
        [ 2799.569315][ T5562] hardirqs last disabled at (1566656): [<ffffffffa30bee38>] __schedule+0x9a8/0x18c0
        [ 2799.570991][ T5562] softirqs last  enabled at (1565224): [<ffffffffa113ad8f>] __irq_exit_rcu+0xef/0x160
        [ 2799.572340][ T5562] softirqs last disabled at (1565219): [<ffffffffa113ad8f>] __irq_exit_rcu+0xef/0x160
        [ 2799.573754][ T5562]
        [ 2799.574131][ T5562] Reported by Kernel Concurrency Sanitizer on:
        [ 2799.574947][ T5562] CPU: 3 PID: 5562 Comm: tee Tainted: G      D           5.19.0-466d9d7ea677-for-next+ #85 89955463945a81b56a449b1f12383cf0d5e6b898
        [ 2799.576954][ T5562] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-2 04/01/2014
        [ 2799.578369][ T5562] ==================================================================

Writer is:

        btrfs_update_delayed_refs_rsv+0xce/0x110:

        btrfs_update_delayed_refs_rsv at fs/btrfs/delayed-ref.c:135
         130            if (btrfs_test_opt(fs_info, FREE_SPACE_TREE))
         131                    num_bytes *= 2;
         132
         133            spin_lock(&delayed_rsv->lock);
         134            delayed_rsv->size += num_bytes;
        >135<           delayed_rsv->full = false;
         136            spin_unlock(&delayed_rsv->lock);
         137            trans->delayed_ref_updates = 0;
         138    }
         139
         140    /**

Reader is:

        btrfs_block_rsv_release+0x5b/0x3f0:

        btrfs_block_rsv_release at fs/btrfs/block-rsv.c:289 (discriminator 1)
         284             * If we are the delayed_rsv then push to the global rsv, otherwise dump
         285             * into the delayed rsv if it is not full.
         286             */
         287            if (block_rsv == delayed_rsv)
         288                    target = global_rsv;
        >289<           else if (block_rsv != global_rsv && !delayed_rsv->full)
         290                    target = delayed_rsv;
         291
         292            if (target && block_rsv->space_info != target->space_info)
         293                    target = NULL;
         294

Another, this time with btrfs_block_rsv_release racing with itself:

        [ 2785.914537][ T3379] ==================================================================
        [ 2785.916300][ T3379] BUG: KCSAN: data-race in btrfs_block_rsv_release / btrfs_block_rsv_release
        [ 2785.918215][ T3379]
        [ 2785.918641][ T3379] read to 0xffff9c80059504c8 of 1 bytes by task 6169 on cpu 2:
        [ 2785.920426][ T3379]  btrfs_block_rsv_release+0x5b/0x3f0
        [ 2785.921671][ T3379]  btrfs_trans_release_metadata+0xae/0xe0
        [ 2785.956873][ T3379]  __btrfs_end_transaction+0x80/0x4f0
        [ 2785.958523][ T3379]  btrfs_end_transaction+0x21/0x40
        [ 2785.959223][ T3379]  btrfs_evict_inode+0x7e9/0x800
        [ 2785.959896][ T3379]  evict+0x1ad/0x330
        [ 2785.960430][ T3379]  iput.part.0+0x2e3/0x410
        [ 2785.964122][ T3379]  iput+0x3c/0x50
        [ 2785.967801][ T3379]  dentry_unlink_inode+0x1f4/0x280
        [ 2785.973195][ T3379]  __dentry_kill+0x21f/0x330
        [ 2785.978567][ T3379]  dput+0x44e/0x810
        [ 2785.983573][ T3379]  do_renameat2+0x4d5/0x900
        [ 2786.037153][ T3379]  __x64_sys_rename+0x67/0x90
        [ 2786.037717][ T3379]  do_syscall_64+0x5f/0xd0
        [ 2786.038242][ T3379]  entry_SYSCALL_64_after_hwframe+0x63/0xcd
        [ 2786.038931][ T3379]
        [ 2786.039208][ T3379] INFO: lockdep is turned off.
        [ 2786.039755][ T3379] irq event stamp: 3250759
        [ 2786.040333][ T3379] hardirqs last  enabled at (3250758): [<ffffffffa30ce452>] _raw_spin_unlock_irq+0x32/0x80
        [ 2786.041701][ T3379] hardirqs last disabled at (3250759): [<ffffffffa30bee38>] __schedule+0x9a8/0x18c0
        [ 2786.043073][ T3379] softirqs last  enabled at (3250361): [<ffffffffa113ad8f>] __irq_exit_rcu+0xef/0x160
        [ 2786.044221][ T3379] softirqs last disabled at (3250352): [<ffffffffa113ad8f>] __irq_exit_rcu+0xef/0x160
        [ 2786.045392][ T3379]
        [ 2786.045712][ T3379] write to 0xffff9c80059504c8 of 1 bytes by task 3379 on cpu 0:
        [ 2786.046658][ T3379]  btrfs_block_rsv_release+0x23e/0x3f0
        [ 2786.047397][ T3379]  btrfs_delayed_refs_rsv_release+0x75/0xd0
        [ 2786.057504][ T3379]  btrfs_cleanup_ref_head_accounting+0x4f/0x140
        [ 2786.058314][ T3379]  __btrfs_run_delayed_refs+0x18c9/0x2080
        [ 2786.059279][ T3379]  btrfs_run_delayed_refs+0xcd/0x310
        [ 2786.061723][ T3379]  btrfs_commit_transaction+0x136/0x1800
        [ 2786.064101][ T3379]  transaction_kthread+0x24c/0x2d0
        [ 2786.066161][ T3379]  kthread+0x1ab/0x1e0
        [ 2786.067969][ T3379]  ret_from_fork+0x22/0x30
        [ 2786.070012][ T3379]
        [ 2786.070391][ T3379] INFO: lockdep is turned off.
        [ 2786.073032][ T3379] irq event stamp: 5460863
        [ 2786.075562][ T3379] hardirqs last  enabled at (5460862): [<ffffffffa30ce51d>] _raw_spin_unlock_irqrestore+0x7d/0xa0
        [ 2786.079952][ T3379] hardirqs last disabled at (5460863): [<ffffffffa30bee38>] __schedule+0x9a8/0x18c0
        [ 2786.084111][ T3379] softirqs last  enabled at (5460221): [<ffffffffa113ad8f>] __irq_exit_rcu+0xef/0x160
        [ 2786.088033][ T3379] softirqs last disabled at (5460214): [<ffffffffa113ad8f>] __irq_exit_rcu+0xef/0x160
        [ 2786.091680][ T3379]
        [ 2786.092464][ T3379] Reported by Kernel Concurrency Sanitizer on:
        [ 2786.095225][ T3379] CPU: 0 PID: 3379 Comm: btrfs-transacti Tainted: G      D           5.19.0-466d9d7ea677-for-next+ #85 89955463945a81b56a449b1f12383cf0d5e6b898
        [ 2786.100076][ T3379] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-2 04/01/2014
        [ 2786.102614][ T3379] ==================================================================
        btrfs_block_rsv_release+0x23e/0x3f0:

        btrfs_block_rsv_release at fs/btrfs/block-rsv.c:121
         116            }
         117            block_rsv->size -= num_bytes;
         118            if (block_rsv->reserved >= block_rsv->size) {
         119                    num_bytes = block_rsv->reserved - block_rsv->size;
         120                    block_rsv->reserved = block_rsv->size;
        >121<                   block_rsv->full = true;
         122            } else {
         123                    num_bytes = 0;
         124            }
         125            if (block_rsv->qgroup_rsv_reserved >= block_rsv->qgroup_rsv_size) {
         126                    qgroup_to_release = block_rsv->qgroup_rsv_reserved -

There's another one in start_transaction. It's basically everywhere, and always looking at some struct's ->full member:

        start_transaction+0x8ce/0xa40:

        start_transaction at fs/btrfs/transaction.c:638 (discriminator 1)
         633                     * worth of delayed refs updates in this trans handle, and
         634                     * refill that amount for whatever is missing in the reserve.
         635                     */
         636                    num_bytes = btrfs_calc_insert_metadata_size(fs_info, num_items);
         637                    if (flush == BTRFS_RESERVE_FLUSH_ALL &&
        >638<                       delayed_refs_rsv->full == 0) {
         639                            delayed_refs_bytes = num_bytes;
         640                            num_bytes <<= 1;
         641                    }
         642
         643                    /*
Signature · 3 lines
Filipe David Manana,

“Whether you think you can, or you think you can't — you're right.”
#2 lore raw
Quoted 7 lines · Filipe Manana
Quoted 4 lines · Zygo Blaxell

Some KCSAN complaints I found while testing for other things...

Here's one related to extent refs:

Quoted 8 lines

It's about the block reserves, nothing to do with extents refs.

These get reported every now and then like here:

https://lore.kernel.org/linux-btrfs/CAAwBoOJDjei5Hnem155N_cJwiEkVwJYvgN-tQrwWbZQGhFU=cA@mail.gmail.com/

It's actually harmless, but if we keep it like this, we'll keep getting reports in the future.

Can we add some kind of annotation so KCSAN understands that? The ->full member would be accessed using a helper when outside of the lock so the annotation can be there.

#3 lore raw
Quoted 12 lines · David Sterba
Quoted 11 lines · Filipe Manana
Quoted 4 lines · Zygo Blaxell

Some KCSAN complaints I found while testing for other things...

Here's one related to extent refs:

It's about the block reserves, nothing to do with extents refs.

These get reported every now and then like here:

Quoted 8 lines
Quoted 4 lines

https://lore.kernel.org/linux-btrfs/CAAwBoOJDjei5Hnem155N_cJwiEkVwJYvgN-tQrwWbZQGhFU=cA@mail.gmail.com/

It's actually harmless, but if we keep it like this, we'll keep getting reports in the future.

Can we add some kind of annotation so KCSAN understands that? The ->full member would be accessed using a helper when outside of the lock so the annotation can be there.

Something like (only compile-tested):

fs/btrfs/block-rsv.c +1 -1
index 6ce704d3bdd2..ec96285357e0 100644
@@ -286,7 +286,7 @@ u64 btrfs_block_rsv_release(struct btrfs_fs_info *fs_info,
 	 */
 	if (block_rsv == delayed_rsv)
 		target = global_rsv;
-	else if (block_rsv != global_rsv && !delayed_rsv->full)
+	else if (block_rsv != global_rsv && !btrfs_block_rsv_full(delayed_rsv))
 		target = delayed_rsv;
 
 	if (target && block_rsv->space_info != target->space_info)
fs/btrfs/block-rsv.h +4 -0
index 0c183709be00..a1f40b88fa82 100644
@@ -91,5 +91,9 @@ static inline void btrfs_unuse_block_rsv(struct btrfs_fs_info *fs_info,
 	btrfs_block_rsv_add_bytes(block_rsv, blocksize, false);
 	btrfs_block_rsv_release(fs_info, block_rsv, 0, NULL);
 }
+static inline bool btrfs_block_rsv_full(const struct btrfs_block_rsv *rsv)
+{
+	return data_race(rsv->full);
+}
 
 #endif /* BTRFS_BLOCK_RSV_H */
fs/btrfs/super.c +1 -1
index a0a2652962ec..eb0ae7e396ef 100644
@@ -2426,7 +2426,7 @@ static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 	 * still can allocate chunks and thus are fine using the currently
 	 * calculated f_bavail.
 	 */
-	if (!mixed && btrfs_block_rsv_full(block_rsv->space_info) &&
+	if (!mixed && block_rsv->space_info->full &&
 	    total_free_meta - thresh < block_rsv->size)
 		buf->f_bavail = 0;
 
fs/btrfs/transaction.c +2 -2
index d9e608935e64..9d7563df81a0 100644
@@ -594,7 +594,7 @@ start_transaction(struct btrfs_root *root, unsigned int num_items,
 		 */
 		num_bytes = btrfs_calc_insert_metadata_size(fs_info, num_items);
 		if (flush == BTRFS_RESERVE_FLUSH_ALL &&
-		    delayed_refs_rsv->full == 0) {
+		    btrfs_block_rsv_full(delayed_refs_rsv) == 0) {
 			delayed_refs_bytes = num_bytes;
 			num_bytes <<= 1;
 		}
@@ -619,7 +619,7 @@ start_transaction(struct btrfs_root *root, unsigned int num_items,
 		if (rsv->space_info->force_alloc)
 			do_chunk_alloc = true;
 	} else if (num_items == 0 && flush == BTRFS_RESERVE_FLUSH_ALL &&
-		   !delayed_refs_rsv->full) {
+		   !btrfs_block_rsv_full(delayed_refs_rsv)) {
 		/*
 		 * Some people call with btrfs_start_transaction(root, 0)
 		 * because they can be throttled, but have some other mechanism
#4 lore raw
Quoted 16 lines · David Sterba
Quoted 14 lines · David Sterba
Quoted 13 lines · Filipe Manana
Quoted 4 lines · Zygo Blaxell

Some KCSAN complaints I found while testing for other things...

Here's one related to extent refs:

It's about the block reserves, nothing to do with extents refs.

These get reported every now and then like here:

https://lore.kernel.org/linux-btrfs/CAAwBoOJDjei5Hnem155N_cJwiEkVwJYvgN-tQrwWbZQGhFU=cA@mail.gmail.com/

Quoted 8 lines
Quoted 6 lines
Quoted 2 lines

It's actually harmless, but if we keep it like this, we'll keep getting reports in the future.

Can we add some kind of annotation so KCSAN understands that? The ->full member would be accessed using a helper when outside of the lock so the annotation can be there.

Something like (only compile-tested):

I was just seeing that at:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/tools/memory-model/Documentation/access-marking.txt

(linked from Documentation/dev-tools/kcsan.rst)

Quoted 55 lines
fs/btrfs/block-rsv.c +1 -1
index 6ce704d3bdd2..ec96285357e0 100644
@@ -286,7 +286,7 @@ u64 btrfs_block_rsv_release(struct btrfs_fs_info *fs_info,
         */
        if (block_rsv == delayed_rsv)
                target = global_rsv;
-       else if (block_rsv != global_rsv && !delayed_rsv->full)
+       else if (block_rsv != global_rsv && !btrfs_block_rsv_full(delayed_rsv))
                target = delayed_rsv;
 
        if (target && block_rsv->space_info != target->space_info)
fs/btrfs/block-rsv.h +4 -0
index 0c183709be00..a1f40b88fa82 100644
@@ -91,5 +91,9 @@ static inline void btrfs_unuse_block_rsv(struct btrfs_fs_info *fs_info,
        btrfs_block_rsv_add_bytes(block_rsv, blocksize, false);
        btrfs_block_rsv_release(fs_info, block_rsv, 0, NULL);
 }
+static inline bool btrfs_block_rsv_full(const struct btrfs_block_rsv *rsv)
+{
+       return data_race(rsv->full);
+}
 
 #endif /* BTRFS_BLOCK_RSV_H */
fs/btrfs/super.c +1 -1
index a0a2652962ec..eb0ae7e396ef 100644
@@ -2426,7 +2426,7 @@ static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
         * still can allocate chunks and thus are fine using the currently
         * calculated f_bavail.
         */
-       if (!mixed && btrfs_block_rsv_full(block_rsv->space_info) &&
+       if (!mixed && block_rsv->space_info->full &&
            total_free_meta - thresh < block_rsv->size)
                buf->f_bavail = 0;
 
fs/btrfs/transaction.c +1 -1
index d9e608935e64..9d7563df81a0 100644
@@ -594,7 +594,7 @@ start_transaction(struct btrfs_root *root, unsigned int num_items,
                 */
                num_bytes = btrfs_calc_insert_metadata_size(fs_info, num_items);
                if (flush == BTRFS_RESERVE_FLUSH_ALL &&
-                   delayed_refs_rsv->full == 0) {
+                   btrfs_block_rsv_full(delayed_refs_rsv) == 0) {
                        delayed_refs_bytes = num_bytes;
                        num_bytes <<= 1;
                }

fs/btrfs/transaction.c start_transaction(struct btrfs_root *root, unsigned int num_items, · line 622

Quoted 5 lines
fs/btrfs/transaction.c +1 -1
@@ -619,7 +619,7 @@ start_transaction(struct btrfs_root *root, unsigned int num_items,
                if (rsv->space_info->force_alloc)
                        do_chunk_alloc = true;
        } else if (num_items == 0 && flush == BTRFS_RESERVE_FLUSH_ALL &&
-                  !delayed_refs_rsv->full) {
+                  !btrfs_block_rsv_full(delayed_refs_rsv)) {

Looks good to me, thanks.

Quoted 3 lines
fs/btrfs/transaction.c +0 -0
                /*
                 * Some people call with btrfs_start_transaction(root, 0)
                 * because they can be throttled, but have some other mechanism