Received: from mail-qv1-f45.google.com (mail-qv1-f45.google.com [209.85.219.45]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id A81B823535E for ; Mon, 18 May 2026 01:32:13 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.219.45 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1779067935; cv=none; b=uTpfytkjbBmHMiKiW+/ff6N1ny36a/ZNu6Lz24eDEnPvVHUfM1GpqwCvdzRRq/bP8J4tis48+Nr9PzLd9g0ig/ruGFsIcvO26LjkNt/xgvdTIhoVpIiQQcxpZi2m6bJ+gIGEg6vKwx4o5qWEYS3vCIY6gGA5pPD4+LfzkL+Bs1Q= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1779067935; c=relaxed/simple; bh=uSxBVG9L5IPEBZomtFI9GNRnEycBuFk/DSraLxzhB6s=; h=From:To:Cc:Subject:Date:Message-ID:MIME-Version; b=hdlVHc+hpOje3oKGsQnKFCndfdYoC0Mrrs8x9gofCzXZBzqLZBqUC/fCVPssq7NcAMYhbQvjWbTyuPvaZYVaeeQq1ZBTtoRIaxQ/099L1xjSEHT1R+5Fwg4/vqeRdm43gqYoWBIzX1kK6edyLrMHNMxIW0Yezc5pkQ/rH5Gx4PM= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=gmail.com; spf=pass smtp.mailfrom=gmail.com; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b=lGcZUUgR; arc=none smtp.client-ip=209.85.219.45 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=gmail.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=gmail.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b="lGcZUUgR" Received: by mail-qv1-f45.google.com with SMTP id 6a1803df08f44-8b8e98fd885so27329206d6.0 for ; Sun, 17 May 2026 18:32:13 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20251104; t=1779067933; x=1779672733; darn=vger.kernel.org; h=content-transfer-encoding:mime-version:message-id:date:subject:cc :to:from:from:to:cc:subject:date:message-id:reply-to; bh=H9l9VOx9k1v+Q7n7JIO0+MFTnyUBN8tJgLVov8FuifQ=; b=lGcZUUgRYCMmVTAiyIPJjzUHm4JOSYy0ttFT/a7sNUK1MIgl9zo1onPoiK60po1n33 mdsiMixUw35lv1auXmI6/rC+jUtpYK1t84vOdYphiw9rHo+pqkjqZqcZFT8G4HUgoso0 PQA++icEU1vVOm9iMsYb0TOrsKE87tGGvqqVFmSqRnq3MJfi4jbLouJKFZV2/EPmQ47C /0ptbstYkixldxoVGxWHIVgMpPJ1UmERvfiVDV/xYO4MGW+B0qIQFbk1ZCl5d9Uk8A43 +HBx7Fgfa0yo84dG5qBwkg/WYdI0ues9L2Ka+HpVvwLyouZkBm+QvZ0xrbSF/nevaeEU GbVw== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20251104; t=1779067933; x=1779672733; h=content-transfer-encoding:mime-version:message-id:date:subject:cc :to:from:x-gm-gg:x-gm-message-state:from:to:cc:subject:date :message-id:reply-to; bh=H9l9VOx9k1v+Q7n7JIO0+MFTnyUBN8tJgLVov8FuifQ=; b=YyP7IXIzfA9kS0i5b2gmP0k748BzgYCl+gNMPZRuLGDWIrdBLQQOsNUK9qpRlRZ9rZ 0PlV2sJntB0DNhC3h1dEzxSihq+QiZiIHz4XMqltJ86TXwJAcOEnyAfpK/rqpBOHD8ur aPtox3rCkogUeR4eC+7RpUODtKSCHLWDoJNPJwFazIalUVPzxRDcb/jNkCbaEAN9K23r GhSjRnQsPBo1I0+1sU3j4URq66/37RrMrHLWheVKODlySy+RaIRjIhmTspKzvG1GLf0u zpPfw0BVjUYw2Ck5LCcFhsAdejnc4OGUKC3gK7H4FaU5mkWC4aTldcNjFQCdvDQF4Z4K UrnA== X-Forwarded-Encrypted: i=1; AFNElJ//Xyc/IYlX4f0sNEefNFLr1RYsYhXfjx/kzxqQcIRhptc3w2jyyRi74nDNosIDZrulfUQosYlhe4I=@vger.kernel.org X-Gm-Message-State: AOJu0Yw0eNalt8DgRHlEslG7OLkhqrzOQlayanZOnIZWvjHIvyK2iBit HIqwNzIKrkHyGPEZW4R2km9A+A5LMLY7qM1rWnRxN6k6wcPbwd5S2CdK X-Gm-Gg: Acq92OHV8/hpXPCMasVdJ3phN12oSEB1WryKMU/qGYijoMWVLrKYzc/aG7ncZxtPbrU ZRvzAQBXzAtCrQLO/6FLTl3sz+9dWShzlqBR0Ax60Tw4CZe9OWuyqi1pYKOAhHoPXqbWe3ejHSb E3Pf04K+WZT7PXFRlxTebTtm9DlhGsQqd7D6C6kLj9kmDWQeBkK+WOOeTjzWvNSzo8nJkcCfXRV /TF0MsPMEtNS+41yvf7CVRXOZT7CEgv4pb4iMi1NHg7E5aojS+7iNCdD0hWh7qCsWXnyilhGE1V HxuLDMDnMk1ZZ6si5qay2rW81h7Ei7WBgtY2ZQ1zKyH+bWsfPaXHW5ebztJOih6aQR8GRs4XS1d zmi1pEObjC5l64TyPP2sYvpU+yhaqJFegeM+xNLEi8bbinssC0wPjEn0IP3pJGfDk1U9RYkHLQN +QOwS15GJiNwdVM0a424AvzlgRl7z/INh4xHE/N/E0Ynrtw0v0i4YbpFhlpATv+SsLc2qimSQPa 28VDgfbom/KrrLva32sBnvH/DPpP3Nf2WSwPr4/LWg= X-Received: by 2002:ad4:4eeb:0:b0:89c:d424:aceb with SMTP id 6a1803df08f44-8ca0f6d639dmr218710626d6.31.1779067932625; Sun, 17 May 2026 18:32:12 -0700 (PDT) Received: from server0.tail6e7dd.ts.net (c-68-48-65-54.hsd1.mi.comcast.net. [68.48.65.54]) by smtp.gmail.com with ESMTPSA id 6a1803df08f44-8ca36086b95sm39111266d6.5.2026.05.17.18.32.11 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Sun, 17 May 2026 18:32:11 -0700 (PDT) From: Michael Bommarito To: Carlos Maiolino Cc: "Darrick J. Wong" , linux-xfs@vger.kernel.org, linux-kernel@vger.kernel.org Subject: [PATCH] xfs: detect cycles in recovered unlinked inode lists Date: Sun, 17 May 2026 21:31:43 -0400 Message-ID: <20260518013143.1532327-1-michael.bommarito@gmail.com> X-Mailer: git-send-email 2.53.0 Precedence: bulk X-Mailing-List: linux-xfs@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Two XFS walkers follow on-disk AGI unlinked bucket lists until the NULLAGINO sentinel and trust the links unconditionally: xlog_recover_iunlink_bucket(), reached at mount time when the log needs recovery, via xlog_recover_finish() -> xlog_recover_process_iunlinks(). xfs_inode_reload_unlinked_bucket(), reached post-mount via xfs_bulkstat_one_int() and other inode-reload paths. A crafted image with an AGI bucket head pointing into a cycle traps either walker forever. With the mount-time walker, the mount(8) syscall itself never returns; the kernel thread is uninterruptible inside the loop, so SIGKILL on the mount process is queued but never delivered. With the bulkstat walker, an XFS_IOC_BULKSTAT call hangs the same way. Reject invalid AG inode numbers, bucket mismatches, and repeated AG inode numbers in both walkers. Mark the AGI sick and return -EFSCORRUPTED so callers handle the corrupt metadata instead of walking the same on-disk list forever. Reproduced on a crafted image whose AGI bucket holds a self-cycle and whose on-disk log is dirty: a stock kernel hangs in the mount syscall indefinitely, while the patched kernel completes log recovery, reports the corruption, and the filesystem shuts down with -EFSCORRUPTED. Fixes: 04755d2e5821b3afbaadd09fe5df58d04de36484 ("xfs: refactor xlog_recover_process_iunlinks()") Fixes: 83771c50e42b92de6740a63e152c96c052d37736 ("xfs: reload entire unlinked bucket lists") Assisted-by: Codex:gpt-5-5-xhigh Assisted-by: Claude:claude-opus-4-7 Signed-off-by: Michael Bommarito --- fs/xfs/xfs_inode.c | 29 +++++++++++++++++++++++++++++ fs/xfs/xfs_log_recover.c | 26 +++++++++++++++++++++++++- 2 files changed, 54 insertions(+), 1 deletion(-) diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c index beaa26ec62da4..f930d5c823c77 100644 --- a/fs/xfs/xfs_inode.c +++ b/fs/xfs/xfs_inode.c @@ -4,6 +4,7 @@ * All Rights Reserved. */ #include +#include #include "xfs_platform.h" #include "xfs_fs.h" @@ -2859,6 +2860,7 @@ xfs_inode_reload_unlinked_bucket( struct xfs_buf *agibp; struct xfs_agi *agi; struct xfs_perag *pag; + struct xarray seen_aginos; xfs_agnumber_t agno = XFS_INO_TO_AGNO(mp, ip->i_ino); xfs_agino_t agino = XFS_INO_TO_AGINO(mp, ip->i_ino); xfs_agino_t prev_agino, next_agino; @@ -2873,6 +2875,8 @@ xfs_inode_reload_unlinked_bucket( if (error) return error; + xa_init(&seen_aginos); + /* * We've taken ILOCK_SHARED and the AGI buffer lock to stabilize the * incore unlinked list pointers for this inode. Check once more to @@ -2897,6 +2901,30 @@ xfs_inode_reload_unlinked_bucket( while (next_agino != NULLAGINO) { struct xfs_inode *next_ip = NULL; + /* + * The on-disk unlinked list is corrupt if it points outside this + * AG, into another bucket, or back to an inode that we already + * saw during this reload walk. + */ + if (!xfs_verify_agino(pag, next_agino) || + next_agino % XFS_AGI_UNLINKED_BUCKETS != bucket) { + xfs_buf_mark_corrupt(agibp); + xfs_ag_mark_sick(pag, XFS_SICK_AG_AGI); + error = -EFSCORRUPTED; + break; + } + + if (xa_load(&seen_aginos, next_agino)) { + xfs_buf_mark_corrupt(agibp); + xfs_ag_mark_sick(pag, XFS_SICK_AG_AGI); + error = -EFSCORRUPTED; + break; + } + error = xa_err(xa_store(&seen_aginos, next_agino, + xa_mk_value(1), GFP_NOFS)); + if (error) + break; + /* Found this caller's inode, set its backlink. */ if (next_agino == agino) { next_ip = ip; @@ -2931,6 +2959,7 @@ xfs_inode_reload_unlinked_bucket( } out_agibp: + xa_destroy(&seen_aginos); xfs_trans_brelse(tp, agibp); /* Should have found this inode somewhere in the iunlinked bucket. */ if (!error && !foundit) diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c index 09e6678ca4878..8ca70bbbfac81 100644 --- a/fs/xfs/xfs_log_recover.c +++ b/fs/xfs/xfs_log_recover.c @@ -3,6 +3,8 @@ * Copyright (c) 2000-2006 Silicon Graphics, Inc. * All Rights Reserved. */ +#include + #include "xfs_platform.h" #include "xfs_fs.h" #include "xfs_shared.h" @@ -28,6 +30,7 @@ #include "xfs_ag.h" #include "xfs_quota.h" #include "xfs_reflink.h" +#include "xfs_health.h" #define BLK_AVG(blk1, blk2) ((blk1+blk2) >> 1) @@ -2726,11 +2729,31 @@ xlog_recover_iunlink_bucket( struct xfs_mount *mp = pag_mount(pag); struct xfs_inode *prev_ip = NULL; struct xfs_inode *ip; + struct xarray seen_aginos; xfs_agino_t prev_agino, agino; int error = 0; + xa_init(&seen_aginos); + agino = be32_to_cpu(agi->agi_unlinked[bucket]); while (agino != NULLAGINO) { + if (!xfs_verify_agino(pag, agino) || + agino % XFS_AGI_UNLINKED_BUCKETS != bucket) { + xfs_ag_mark_sick(pag, XFS_SICK_AG_AGI); + error = -EFSCORRUPTED; + break; + } + + if (xa_load(&seen_aginos, agino)) { + xfs_ag_mark_sick(pag, XFS_SICK_AG_AGI); + error = -EFSCORRUPTED; + break; + } + error = xa_err(xa_store(&seen_aginos, agino, xa_mk_value(1), + GFP_NOFS)); + if (error) + break; + error = xfs_iget(mp, NULL, xfs_agino_to_ino(pag, agino), 0, 0, &ip); if (error) @@ -2771,8 +2794,9 @@ xlog_recover_iunlink_bucket( error2 = xfs_inodegc_flush(mp); if (error2 && !error) - return error2; + error = error2; } + xa_destroy(&seen_aginos); return error; } -- 2.53.0