* [RFC 1/1 linux-next] udf: allow implicit blocksize specification during mount
@ 2017-01-18 18:39 Fabian Frederick
2017-01-20 10:12 ` Jan Kara
0 siblings, 1 reply; 2+ messages in thread
From: Fabian Frederick @ 2017-01-18 18:39 UTC (permalink / raw)
To: Jan Kara; +Cc: linux-kernel, linux-fsdevel, fabf
udf_fill_super() used udf_parse_options() to flag UDF_FLAG_BLOCKSIZE_SET
when blocksize was specified otherwise used 512 bytes
(bdev_logical_block_size) and 2048 bytes (UDF_DEFAULT_BLOCKSIZE)
IOW both 1024 and 4096 specifications were required or resulted in
"mount: wrong fs type, bad option, bad superblock on /dev/loop1"
This patch loops through different block values but also updates
udf_load_vrs() to return -EINVAL instead of 0 when udf_check_vsd()
fails (and uopt->novrs = 0).
The later being the reason for the RFC; we have that case when mounting
a 4kb blocksize against other values but maybe VRS is not mandatory
there ?
Tested with 512, 1024, 2048 and 4096 blocksize
Reported-by: Jan Kara <jack@suse.com>
Signed-off-by: Fabian Frederick <fabf@skynet.be>
---
fs/udf/super.c | 22 +++++++++++++---------
1 file changed, 13 insertions(+), 9 deletions(-)
diff --git a/fs/udf/super.c b/fs/udf/super.c
index 967ad87..078a144 100644
--- a/fs/udf/super.c
+++ b/fs/udf/super.c
@@ -1957,7 +1957,7 @@ static int udf_load_vrs(struct super_block *sb, struct udf_options *uopt,
if (!nsr_off) {
if (!silent)
udf_warn(sb, "No VRS found\n");
- return 0;
+ return -EINVAL;
}
if (nsr_off == -1)
udf_debug("Failed to read sector at offset %d. "
@@ -2161,15 +2161,19 @@ static int udf_fill_super(struct super_block *sb, void *options, int silent)
ret = udf_load_vrs(sb, &uopt, silent, &fileset);
} else {
uopt.blocksize = bdev_logical_block_size(sb->s_bdev);
- ret = udf_load_vrs(sb, &uopt, silent, &fileset);
- if (ret == -EAGAIN && uopt.blocksize != UDF_DEFAULT_BLOCKSIZE) {
- if (!silent)
- pr_notice("Rescanning with blocksize %d\n",
- UDF_DEFAULT_BLOCKSIZE);
- brelse(sbi->s_lvid_bh);
- sbi->s_lvid_bh = NULL;
- uopt.blocksize = UDF_DEFAULT_BLOCKSIZE;
+ while (uopt.blocksize <= 4096) {
ret = udf_load_vrs(sb, &uopt, silent, &fileset);
+ if (ret < 0) {
+ if (!silent) {
+ pr_notice("Scanning with blocksize %d failed\n",
+ uopt.blocksize);
+ }
+ brelse(sbi->s_lvid_bh);
+ sbi->s_lvid_bh = NULL;
+ } else
+ break;
+
+ uopt.blocksize <<= 1;
}
}
if (ret < 0) {
--
2.9.3
^ permalink raw reply related [flat|nested] 2+ messages in thread
* Re: [RFC 1/1 linux-next] udf: allow implicit blocksize specification during mount
2017-01-18 18:39 [RFC 1/1 linux-next] udf: allow implicit blocksize specification during mount Fabian Frederick
@ 2017-01-20 10:12 ` Jan Kara
0 siblings, 0 replies; 2+ messages in thread
From: Jan Kara @ 2017-01-20 10:12 UTC (permalink / raw)
To: Fabian Frederick; +Cc: Jan Kara, linux-kernel, linux-fsdevel
On Wed 18-01-17 19:39:35, Fabian Frederick wrote:
> udf_fill_super() used udf_parse_options() to flag UDF_FLAG_BLOCKSIZE_SET
> when blocksize was specified otherwise used 512 bytes
> (bdev_logical_block_size) and 2048 bytes (UDF_DEFAULT_BLOCKSIZE)
> IOW both 1024 and 4096 specifications were required or resulted in
>
> "mount: wrong fs type, bad option, bad superblock on /dev/loop1"
>
> This patch loops through different block values but also updates
> udf_load_vrs() to return -EINVAL instead of 0 when udf_check_vsd()
> fails (and uopt->novrs = 0).
> The later being the reason for the RFC; we have that case when mounting
> a 4kb blocksize against other values but maybe VRS is not mandatory
> there ?
>
> Tested with 512, 1024, 2048 and 4096 blocksize
>
> Reported-by: Jan Kara <jack@suse.com>
> Signed-off-by: Fabian Frederick <fabf@skynet.be>
Thanks for the patch. It looks good to me. I'll test it a bit and pick it
up.
Honza
> ---
> fs/udf/super.c | 22 +++++++++++++---------
> 1 file changed, 13 insertions(+), 9 deletions(-)
>
> diff --git a/fs/udf/super.c b/fs/udf/super.c
> index 967ad87..078a144 100644
> --- a/fs/udf/super.c
> +++ b/fs/udf/super.c
> @@ -1957,7 +1957,7 @@ static int udf_load_vrs(struct super_block *sb, struct udf_options *uopt,
> if (!nsr_off) {
> if (!silent)
> udf_warn(sb, "No VRS found\n");
> - return 0;
> + return -EINVAL;
> }
> if (nsr_off == -1)
> udf_debug("Failed to read sector at offset %d. "
> @@ -2161,15 +2161,19 @@ static int udf_fill_super(struct super_block *sb, void *options, int silent)
> ret = udf_load_vrs(sb, &uopt, silent, &fileset);
> } else {
> uopt.blocksize = bdev_logical_block_size(sb->s_bdev);
> - ret = udf_load_vrs(sb, &uopt, silent, &fileset);
> - if (ret == -EAGAIN && uopt.blocksize != UDF_DEFAULT_BLOCKSIZE) {
> - if (!silent)
> - pr_notice("Rescanning with blocksize %d\n",
> - UDF_DEFAULT_BLOCKSIZE);
> - brelse(sbi->s_lvid_bh);
> - sbi->s_lvid_bh = NULL;
> - uopt.blocksize = UDF_DEFAULT_BLOCKSIZE;
> + while (uopt.blocksize <= 4096) {
> ret = udf_load_vrs(sb, &uopt, silent, &fileset);
> + if (ret < 0) {
> + if (!silent) {
> + pr_notice("Scanning with blocksize %d failed\n",
> + uopt.blocksize);
> + }
> + brelse(sbi->s_lvid_bh);
> + sbi->s_lvid_bh = NULL;
> + } else
> + break;
> +
> + uopt.blocksize <<= 1;
> }
> }
> if (ret < 0) {
> --
> 2.9.3
>
--
Jan Kara <jack@suse.com>
SUSE Labs, CR
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2017-01-20 10:13 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-01-18 18:39 [RFC 1/1 linux-next] udf: allow implicit blocksize specification during mount Fabian Frederick
2017-01-20 10:12 ` Jan Kara
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).