2008-10-10 22:14:10

by Eric Sandeen

Subject: [PATCH] e2fsprogs: exit from preenhalt if IO errors were encountered

Resolves RH Bug 465679- e2fsck -p crashes with read-only DM device (e.g. logical volume)

If a block device is unwritable, e2fsck -p gets into an infinite loop
trying to preenhalt, close & flush the fs, which tries to flush the cache,
which gets a write error and calls preenhalt which tries to close & flush
the fs ... ad infinitum.

Per Ted's suggestion just flag the ctx as "exiting" and short-circuit
the infinite loop.

Tested by running e2fsck -p on a block device set read-only by BLKROSET.

Signed-off-by: Eric Sandeen <[email protected]>

Index: e2fsprogs/e2fsck/e2fsck.h
===================================================================
--- e2fsprogs.orig/e2fsck/e2fsck.h
+++ e2fsprogs/e2fsck/e2fsck.h
@@ -174,6 +174,7 @@ struct resource_track {
#define E2F_FLAG_RESTARTED 0x0200 /* E2fsck has been restarted */
#define E2F_FLAG_RESIZE_INODE 0x0400 /* Request to recreate resize inode */
#define E2F_FLAG_GOT_DEVSIZE 0x0800 /* Device size has been fetched */
+#define E2F_FLAG_EXITING 0x1000 /* E2fsck exiting due to errors */

/*
* Defines for indicating the e2fsck pass number
Index: e2fsprogs/e2fsck/ehandler.c
===================================================================
--- e2fsprogs.orig/e2fsck/ehandler.c
+++ e2fsprogs/e2fsck/ehandler.c
@@ -33,7 +33,8 @@ static errcode_t e2fsck_handle_read_erro
e2fsck_t ctx;

ctx = (e2fsck_t) fs->priv_data;
-
+ if (ctx->flags & E2F_FLAG_EXITING)
+ return 0;
/*
* If more than one block was read, try reading each block
* separately. We could use the actual bytes read to figure
@@ -79,6 +80,8 @@ static errcode_t e2fsck_handle_write_err
e2fsck_t ctx;

ctx = (e2fsck_t) fs->priv_data;
+ if (ctx->flags & E2F_FLAG_EXITING)
+ return 0;

/*
* If more than one block was written, try writing each block
Index: e2fsprogs/e2fsck/util.c
===================================================================
--- e2fsprogs.orig/e2fsck/util.c
+++ e2fsprogs/e2fsck/util.c
@@ -257,6 +257,7 @@ void preenhalt(e2fsck_t ctx)
fprintf(stderr, _("\n\n%s: UNEXPECTED INCONSISTENCY; "
"RUN fsck MANUALLY.\n\t(i.e., without -a or -p options)\n"),
ctx->device_name);
+ ctx->flags |= E2F_FLAG_EXITING;
if (fs != NULL) {
fs->super->s_state |= EXT2_ERROR_FS;
ext2fs_mark_super_dirty(fs);

2008-10-10 22:39:03

by Eric Sandeen

Subject: Re: [PATCH] e2fsprogs: exit from preenhalt if IO errors were encountered

Eric Sandeen wrote:
> Resolves RH Bug 465679- e2fsck -p crashes with read-only DM device (e.g. logical volume)
>
> If a block device is unwritable, e2fsck -p gets into an infinite loop
> trying to preenhalt, close & flush the fs, which tries to flush the cache,
> which gets a write error and calls preenhalt which tries to close & flush
> the fs ... ad infinitum.
>
> Per Ted's suggestion just flag the ctx as "exiting" and short-circuit
> the infinite loop.
>
> Tested by running e2fsck -p on a block device set read-only by BLKROSET.

Oh, thanks to Vlado Potisk for reporting this.

-Eric