linux-ext4.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] ext4, jbd2: add REQ_FUA flag when recording an error flag
@ 2015-06-23  2:19 Daeho Jeong
       [not found] ` <CAFKwG1g2_+XsdT0pQcNTEYg4SAxDtewYnMsJcS7Hop_FMoEQ5g@mail.gmail.com>
  0 siblings, 1 reply; 2+ messages in thread
From: Daeho Jeong @ 2015-06-23  2:19 UTC (permalink / raw)
  To: linux-ext4

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=utf-8, Size: 2119 bytes --]

When an error condition is detected, an error status should be recorded into
superblocks of EXT4 or JBD2. However, the write request is submitted now
without REQ_FUA flag, even in "barrier=1" mode, which is followed by
panic() function in "errors=panic" mode. On mobile devices which make
whole system reset as soon as kernel panic occurs, this write request
containing an error flag will disappear just from storage cache without
written to the physical cells. Therefore, when next start, even forever,
the error flag cannot be shown in both superblocks, and e2fsck cannot fix
the filesystem problems automatically, unless e2fsck is executed in
force checking mode.

Signed-off-by: Daeho Jeong <daeho.jeong@samsung.com>
---
 fs/ext4/super.c   |    6 +++++-
 fs/jbd2/journal.c |    2 +-
 2 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index ca9d4a2..4194328 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -4588,6 +4588,7 @@ static int ext4_commit_super(struct super_block *sb, int sync)
 {
  struct ext4_super_block *es = EXT4_SB(sb)->s_es;
  struct buffer_head *sbh = EXT4_SB(sb)->s_sbh;
+ journal_t *journal = EXT4_SB(sb)->s_journal;
  int error = 0;
 
  if (!sbh)
@@ -4638,7 +4639,10 @@ static int ext4_commit_super(struct super_block *sb, int sync)
  ext4_superblock_csum_set(sb);
  mark_buffer_dirty(sbh);
  if (sync) {
-  error = sync_dirty_buffer(sbh);
+  if (journal->j_flags & JBD2_BARRIER)
+   error = __sync_dirty_buffer(sbh, WRITE_FUA);
+  else
+   error = sync_dirty_buffer(sbh);
   if (error)
    return error;
 
diff --git a/fs/jbd2/journal.c b/fs/jbd2/journal.c
index b96bd80..5b9693d 100644
--- a/fs/jbd2/journal.c
+++ b/fs/jbd2/journal.c
@@ -1455,7 +1455,7 @@ void jbd2_journal_update_sb_errno(journal_t *journal)
  sb->s_errno    = cpu_to_be32(journal->j_errno);
  read_unlock(&journal->j_state_lock);
 
- jbd2_write_superblock(journal, WRITE_SYNC);
+ jbd2_write_superblock(journal, WRITE_FUA);
 }
 EXPORT_SYMBOL(jbd2_journal_update_sb_errno);
 
N‹§²æìr¸›yúèšØb²X¬¶Ç§vØ^–)Þº{.nÇ+‰·¥Š{±{^[xŠ

^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH] ext4, jbd2: add REQ_FUA flag when recording an error flag
       [not found] ` <CAFKwG1g2_+XsdT0pQcNTEYg4SAxDtewYnMsJcS7Hop_FMoEQ5g@mail.gmail.com>
@ 2015-07-23 13:48   ` Theodore Ts'o
  0 siblings, 0 replies; 2+ messages in thread
From: Theodore Ts'o @ 2015-07-23 13:48 UTC (permalink / raw)
  To: Daeho Jeong; +Cc: daeho.jeong, linux-ext4

> @@ -4638,7 +4639,10 @@ static int ext4_commit_super(struct super_block *sb,
> int sync)
>   ext4_superblock_csum_set(sb);
>   mark_buffer_dirty(sbh);
>   if (sync) {
> -  error = sync_dirty_buffer(sbh);
> +  if (journal->j_flags & JBD2_BARRIER)
> +   error = __sync_dirty_buffer(sbh, WRITE_FUA);
> +  else
> +   error = sync_dirty_buffer(sbh);
>    if (error)
>     return error;

The problem with this is that the journal structure will not be
present in no-journal mode, so it will result in a NULL dereference.
So I've changed this to use test_opt(sb, BARRIER) instead.

   		     	    		 	  - Ted
						  
commit 564bc402526e437729ecafe3c3511f7cab9f0327
Author: Daeho Jeong <daeho.jeong@samsung.com>
Date:   Thu Jul 23 09:46:11 2015 -0400

    ext4, jbd2: add REQ_FUA flag when recording an error in the superblock
    
    When an error condition is detected, an error status should be recorded into
    superblocks of EXT4 or JBD2. However, the write request is submitted now
    without REQ_FUA flag, even in "barrier=1" mode, which is followed by
    panic() function in "errors=panic" mode. On mobile devices which make
    whole system reset as soon as kernel panic occurs, this write request
    containing an error flag will disappear just from storage cache without
    written to the physical cells. Therefore, when next start, even forever,
    the error flag cannot be shown in both superblocks, and e2fsck cannot fix
    the filesystem problems automatically, unless e2fsck is executed in
    force checking mode.
    
    [ Changed use test_opt(sb, BARRIER) of checking the journal flags -- TYT ]
    
    Signed-off-by: Daeho Jeong <daeho.jeong@samsung.com>
    Signed-off-by: Theodore Ts'o <tytso@mit.edu>

diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index 14909cd..a51db9c 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -4667,7 +4667,8 @@ static int ext4_commit_super(struct super_block *sb, int sync)
 	ext4_superblock_csum_set(sb);
 	mark_buffer_dirty(sbh);
 	if (sync) {
-		error = sync_dirty_buffer(sbh);
+		error = __sync_dirty_buffer(sbh,
+			test_opt(sb, BARRIER) ? WRITE_FUA : WRITE_SYNC);
 		if (error)
 			return error;
 
diff --git a/fs/jbd2/journal.c b/fs/jbd2/journal.c
index 4ff3fad..fe1b4bd 100644
--- a/fs/jbd2/journal.c
+++ b/fs/jbd2/journal.c
@@ -1456,7 +1456,7 @@ void jbd2_journal_update_sb_errno(journal_t *journal)
 	sb->s_errno    = cpu_to_be32(journal->j_errno);
 	read_unlock(&journal->j_state_lock);
 
-	jbd2_write_superblock(journal, WRITE_SYNC);
+	jbd2_write_superblock(journal, WRITE_FUA);
 }
 EXPORT_SYMBOL(jbd2_journal_update_sb_errno);
 

^ permalink raw reply related	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2015-07-23 13:48 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2015-06-23  2:19 [PATCH] ext4, jbd2: add REQ_FUA flag when recording an error flag Daeho Jeong
     [not found] ` <CAFKwG1g2_+XsdT0pQcNTEYg4SAxDtewYnMsJcS7Hop_FMoEQ5g@mail.gmail.com>
2015-07-23 13:48   ` Theodore Ts'o

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).