All of lore.kernel.org
 help / color / mirror / Atom feed
* + nilfs2-do-not-propagate-enoent-error-from-sufile-during-gc.patch added to mm-nonmm-unstable branch
@ 2024-08-21 21:24 Andrew Morton
  0 siblings, 0 replies; only message in thread
From: Andrew Morton @ 2024-08-21 21:24 UTC (permalink / raw)
  To: mm-commits, konishi.ryusuke, akpm


The patch titled
     Subject: nilfs2: do not propagate ENOENT error from sufile during GC
has been added to the -mm mm-nonmm-unstable branch.  Its filename is
     nilfs2-do-not-propagate-enoent-error-from-sufile-during-gc.patch

This patch will shortly appear at
     https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patches/nilfs2-do-not-propagate-enoent-error-from-sufile-during-gc.patch

This patch will later appear in the mm-nonmm-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: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Subject: nilfs2: do not propagate ENOENT error from sufile during GC
Date: Thu, 22 Aug 2024 00:46:26 +0900

nilfs_sufile_freev(), which is used to free segments in GC, aborts with
-ENOENT if the target segment usage is on a hole block.

This error only occurs if one of the segment numbers to be freed passed by
the GC ioctl is invalid, so return -EINVAL instead.

To avoid impairing readability, introduce a wrapper function that
encapsulates error handling including the error code conversion (and error
message output).

Link: https://lkml.kernel.org/r/20240821154627.11848-5-konishi.ryusuke@gmail.com
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 fs/nilfs2/segment.c |   64 ++++++++++++++++++++++++++++++++++++------
 1 file changed, 56 insertions(+), 8 deletions(-)

--- a/fs/nilfs2/segment.c~nilfs2-do-not-propagate-enoent-error-from-sufile-during-gc
+++ a/fs/nilfs2/segment.c
@@ -1102,12 +1102,64 @@ static int nilfs_segctor_scan_file_dsync
 	return err;
 }
 
+/**
+ * nilfs_free_segments - free the segments given by an array of segment numbers
+ * @nilfs:   nilfs object
+ * @segnumv: array of segment numbers to be freed
+ * @nsegs:   number of segments to be freed in @segnumv
+ *
+ * nilfs_free_segments() wraps nilfs_sufile_freev() and
+ * nilfs_sufile_cancel_freev(), and edits the segment usage metadata file
+ * (sufile) to free all segments given by @segnumv and @nsegs at once.  If
+ * it fails midway, it cancels the changes so that none of the segments are
+ * freed.  If @nsegs is 0, this function does nothing.
+ *
+ * The freeing of segments is not finalized until the writing of a log with
+ * a super root block containing this sufile change is complete, and it can
+ * be canceled with nilfs_sufile_cancel_freev() until then.
+ *
+ * Return: 0 on success, or the following negative error code on failure.
+ * * %-EINVAL	- Invalid segment number.
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+static int nilfs_free_segments(struct the_nilfs *nilfs, __u64 *segnumv,
+			       size_t nsegs)
+{
+	size_t ndone;
+	int ret;
+
+	if (!nsegs)
+		return 0;
+
+	ret = nilfs_sufile_freev(nilfs->ns_sufile, segnumv, nsegs, &ndone);
+	if (unlikely(ret)) {
+		nilfs_sufile_cancel_freev(nilfs->ns_sufile, segnumv, ndone,
+					  NULL);
+		/*
+		 * If a segment usage of the segments to be freed is in a
+		 * hole block, nilfs_sufile_freev() will return -ENOENT.
+		 * In this case, -EINVAL should be returned to the caller
+		 * since there is something wrong with the given segment
+		 * number array.  This error can only occur during GC, so
+		 * there is no need to worry about it propagating to other
+		 * callers (such as fsync).
+		 */
+		if (ret == -ENOENT) {
+			nilfs_err(nilfs->ns_sb,
+				  "The segment usage entry %llu to be freed is invalid (in a hole)",
+				  (unsigned long long)segnumv[ndone]);
+			ret = -EINVAL;
+		}
+	}
+	return ret;
+}
+
 static int nilfs_segctor_collect_blocks(struct nilfs_sc_info *sci, int mode)
 {
 	struct the_nilfs *nilfs = sci->sc_super->s_fs_info;
 	struct list_head *head;
 	struct nilfs_inode_info *ii;
-	size_t ndone;
 	int err = 0;
 
 	switch (nilfs_sc_cstage_get(sci)) {
@@ -1201,14 +1253,10 @@ static int nilfs_segctor_collect_blocks(
 		nilfs_sc_cstage_inc(sci);
 		fallthrough;
 	case NILFS_ST_SUFILE:
-		err = nilfs_sufile_freev(nilfs->ns_sufile, sci->sc_freesegs,
-					 sci->sc_nfreesegs, &ndone);
-		if (unlikely(err)) {
-			nilfs_sufile_cancel_freev(nilfs->ns_sufile,
-						  sci->sc_freesegs, ndone,
-						  NULL);
+		err = nilfs_free_segments(nilfs, sci->sc_freesegs,
+					  sci->sc_nfreesegs);
+		if (unlikely(err))
 			break;
-		}
 		sci->sc_stage.flags |= NILFS_CF_SUFREED;
 
 		err = nilfs_segctor_scan_file(sci, nilfs->ns_sufile,
_

Patches currently in -mm which might be from konishi.ryusuke@gmail.com are

nilfs2-protect-references-to-superblock-parameters-exposed-in-sysfs.patch
nilfs2-fix-missing-cleanup-on-rollforward-recovery-error.patch
nilfs2-fix-state-management-in-error-path-of-log-writing-function.patch
nilfs2-add-support-for-fs_ioc_getuuid.patch
nilfs2-add-support-for-fs_ioc_getfssysfspath.patch
nilfs2-add-support-for-fs_ioc_getfslabel.patch
nilfs2-add-support-for-fs_ioc_setfslabel.patch
nilfs2-do-not-output-warnings-when-clearing-dirty-buffers.patch
nilfs2-add-missing-argument-description-for-__nilfs_error.patch
nilfs2-add-missing-argument-descriptions-for-ioctl-related-helpers.patch
nilfs2-improve-kernel-doc-comments-for-b-tree-node-helpers.patch
nilfs2-fix-incorrect-kernel-doc-declaration-of-nilfs_palloc_req-structure.patch
nilfs2-add-missing-description-of-nilfs_btree_path-structure.patch
nilfs2-describe-the-members-of-nilfs_bmap_operations-structure.patch
nilfs2-fix-inconsistencies-in-kernel-doc-comments-in-segmenth.patch
nilfs2-fix-missing-initial-short-descriptions-of-kernel-doc-comments.patch
nilfs2-treat-missing-sufile-header-block-as-metadata-corruption.patch
nilfs2-treat-missing-cpfile-header-block-as-metadata-corruption.patch
nilfs2-do-not-propagate-enoent-error-from-sufile-during-recovery.patch
nilfs2-do-not-propagate-enoent-error-from-sufile-during-gc.patch
nilfs2-do-not-propagate-enoent-error-from-nilfs_sufile_mark_dirty.patch


^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2024-08-21 21:24 UTC | newest]

Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2024-08-21 21:24 + nilfs2-do-not-propagate-enoent-error-from-sufile-during-gc.patch added to mm-nonmm-unstable branch Andrew Morton

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.