* [PATCH] xfs_repair: stop using ->data_entry_p()
@ 2020-01-17 23:17 Eric Sandeen
2020-01-18 4:39 ` Darrick J. Wong
2020-01-20 8:58 ` Christoph Hellwig
0 siblings, 2 replies; 6+ messages in thread
From: Eric Sandeen @ 2020-01-17 23:17 UTC (permalink / raw)
To: linux-xfs, Darrick J. Wong
The ->data_entry_p() op went away in v5.5 kernelspace, so rework
xfs_repair to use ->data_entry_offset instead, in preparation
for the v5.5 libxfs backport.
This could later be cleaned up to use offsets as was done
in kernel commit 8073af5153c for example.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
---
I'll munge this patch in mid-libxfs-sync, just before the
->data_entry_p removal patch.
diff --git a/repair/dir2.c b/repair/dir2.c
index 4ac0084e..2494f3c4 100644
--- a/repair/dir2.c
+++ b/repair/dir2.c
@@ -580,7 +580,7 @@ process_dir2_data(
d = bp->b_addr;
bf = M_DIROPS(mp)->data_bestfree_p(d);
- ptr = (char *)M_DIROPS(mp)->data_entry_p(d);
+ ptr = (char *)d + M_DIROPS(mp)->data_entry_offset;
badbest = lastfree = freeseen = 0;
if (be16_to_cpu(bf[0].length) == 0) {
badbest |= be16_to_cpu(bf[0].offset) != 0;
@@ -646,7 +646,7 @@ process_dir2_data(
do_warn(_("\twould junk block\n"));
return 1;
}
- ptr = (char *)M_DIROPS(mp)->data_entry_p(d);
+ ptr = (char *)d + M_DIROPS(mp)->data_entry_offset;
/*
* Process the entries now.
*/
diff --git a/repair/phase6.c b/repair/phase6.c
index 91d208a6..d61b2ae7 100644
--- a/repair/phase6.c
+++ b/repair/phase6.c
@@ -1530,7 +1530,7 @@ longform_dir2_entry_check_data(
bp = *bpp;
d = bp->b_addr;
- ptr = (char *)M_DIROPS(mp)->data_entry_p(d);
+ ptr = (char *)d + M_DIROPS(mp)->data_entry_offset;
nbad = 0;
needscan = needlog = 0;
junkit = 0;
@@ -1590,7 +1590,7 @@ longform_dir2_entry_check_data(
break;
/* check for block with no data entries */
- if ((ptr == (char *)M_DIROPS(mp)->data_entry_p(d)) &&
+ if ((ptr == (char *)d + M_DIROPS(mp)->data_entry_offset) &&
(ptr + be16_to_cpu(dup->length) >= endptr)) {
junkit = 1;
*num_illegal += 1;
@@ -1659,7 +1659,7 @@ longform_dir2_entry_check_data(
do_warn(_("would fix magic # to %#x\n"), wantmagic);
}
lastfree = 0;
- ptr = (char *)M_DIROPS(mp)->data_entry_p(d);
+ ptr = (char *)d + M_DIROPS(mp)->data_entry_offset;
/*
* look at each entry. reference inode pointed to by each
* entry in the incore inode tree.
@@ -1834,7 +1834,7 @@ longform_dir2_entry_check_data(
(dep->name[0] == '.' && dep->namelen == 1));
add_inode_ref(current_irec, current_ino_offset);
if (da_bno != 0 ||
- dep != M_DIROPS(mp)->data_entry_p(d)) {
+ dep != (void *)d + M_DIROPS(mp)->data_entry_offset) {
/* "." should be the first entry */
nbad++;
if (entry_junked(
^ permalink raw reply related [flat|nested] 6+ messages in thread* Re: [PATCH] xfs_repair: stop using ->data_entry_p()
2020-01-17 23:17 [PATCH] xfs_repair: stop using ->data_entry_p() Eric Sandeen
@ 2020-01-18 4:39 ` Darrick J. Wong
2020-01-20 14:29 ` Eric Sandeen
2020-01-20 8:58 ` Christoph Hellwig
1 sibling, 1 reply; 6+ messages in thread
From: Darrick J. Wong @ 2020-01-18 4:39 UTC (permalink / raw)
To: Eric Sandeen; +Cc: linux-xfs
On Fri, Jan 17, 2020 at 05:17:11PM -0600, Eric Sandeen wrote:
> The ->data_entry_p() op went away in v5.5 kernelspace, so rework
> xfs_repair to use ->data_entry_offset instead, in preparation
> for the v5.5 libxfs backport.
>
> This could later be cleaned up to use offsets as was done
> in kernel commit 8073af5153c for example.
See, now that you've said that, I start wondering why not do that?
:D
> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
> ---
>
> I'll munge this patch in mid-libxfs-sync, just before the
> ->data_entry_p removal patch.
>
> diff --git a/repair/dir2.c b/repair/dir2.c
> index 4ac0084e..2494f3c4 100644
> --- a/repair/dir2.c
> +++ b/repair/dir2.c
> @@ -580,7 +580,7 @@ process_dir2_data(
>
> d = bp->b_addr;
> bf = M_DIROPS(mp)->data_bestfree_p(d);
> - ptr = (char *)M_DIROPS(mp)->data_entry_p(d);
> + ptr = (char *)d + M_DIROPS(mp)->data_entry_offset;
> badbest = lastfree = freeseen = 0;
> if (be16_to_cpu(bf[0].length) == 0) {
> badbest |= be16_to_cpu(bf[0].offset) != 0;
> @@ -646,7 +646,7 @@ process_dir2_data(
> do_warn(_("\twould junk block\n"));
> return 1;
> }
> - ptr = (char *)M_DIROPS(mp)->data_entry_p(d);
> + ptr = (char *)d + M_DIROPS(mp)->data_entry_offset;
> /*
> * Process the entries now.
> */
> diff --git a/repair/phase6.c b/repair/phase6.c
> index 91d208a6..d61b2ae7 100644
> --- a/repair/phase6.c
> +++ b/repair/phase6.c
> @@ -1530,7 +1530,7 @@ longform_dir2_entry_check_data(
>
> bp = *bpp;
> d = bp->b_addr;
> - ptr = (char *)M_DIROPS(mp)->data_entry_p(d);
> + ptr = (char *)d + M_DIROPS(mp)->data_entry_offset;
> nbad = 0;
> needscan = needlog = 0;
> junkit = 0;
> @@ -1590,7 +1590,7 @@ longform_dir2_entry_check_data(
> break;
>
> /* check for block with no data entries */
> - if ((ptr == (char *)M_DIROPS(mp)->data_entry_p(d)) &&
> + if ((ptr == (char *)d + M_DIROPS(mp)->data_entry_offset) &&
> (ptr + be16_to_cpu(dup->length) >= endptr)) {
> junkit = 1;
> *num_illegal += 1;
> @@ -1659,7 +1659,7 @@ longform_dir2_entry_check_data(
> do_warn(_("would fix magic # to %#x\n"), wantmagic);
> }
> lastfree = 0;
> - ptr = (char *)M_DIROPS(mp)->data_entry_p(d);
> + ptr = (char *)d + M_DIROPS(mp)->data_entry_offset;
> /*
> * look at each entry. reference inode pointed to by each
> * entry in the incore inode tree.
> @@ -1834,7 +1834,7 @@ longform_dir2_entry_check_data(
> (dep->name[0] == '.' && dep->namelen == 1));
> add_inode_ref(current_irec, current_ino_offset);
> if (da_bno != 0 ||
> - dep != M_DIROPS(mp)->data_entry_p(d)) {
> + dep != (void *)d + M_DIROPS(mp)->data_entry_offset) {
Er.... void pointer arithmetic?
(Though I really do wish the original author of the kernel patchset had
supplied a xfsprogs port of the kernel patches so we maintainers don't
get stuck doing all the porting work...)
--D
> /* "." should be the first entry */
> nbad++;
> if (entry_junked(
>
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] xfs_repair: stop using ->data_entry_p()
2020-01-18 4:39 ` Darrick J. Wong
@ 2020-01-20 14:29 ` Eric Sandeen
2020-01-21 18:37 ` Darrick J. Wong
0 siblings, 1 reply; 6+ messages in thread
From: Eric Sandeen @ 2020-01-20 14:29 UTC (permalink / raw)
To: Darrick J. Wong; +Cc: linux-xfs
On 1/17/20 10:39 PM, Darrick J. Wong wrote:
> On Fri, Jan 17, 2020 at 05:17:11PM -0600, Eric Sandeen wrote:
>> The ->data_entry_p() op went away in v5.5 kernelspace, so rework
>> xfs_repair to use ->data_entry_offset instead, in preparation
>> for the v5.5 libxfs backport.
>>
>> This could later be cleaned up to use offsets as was done
>> in kernel commit 8073af5153c for example.
>
> See, now that you've said that, I start wondering why not do that?
Because this is the fast/safe path to getting the libxfs merge done IMHO ;)
...
>> @@ -1834,7 +1834,7 @@ longform_dir2_entry_check_data(
>> (dep->name[0] == '.' && dep->namelen == 1));
>> add_inode_ref(current_irec, current_ino_offset);
>> if (da_bno != 0 ||
>> - dep != M_DIROPS(mp)->data_entry_p(d)) {
>> + dep != (void *)d + M_DIROPS(mp)->data_entry_offset) {
>
> Er.... void pointer arithmetic?
er, let me take another look at that.
-eric
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] xfs_repair: stop using ->data_entry_p()
2020-01-20 14:29 ` Eric Sandeen
@ 2020-01-21 18:37 ` Darrick J. Wong
2020-01-22 0:41 ` Darrick J. Wong
0 siblings, 1 reply; 6+ messages in thread
From: Darrick J. Wong @ 2020-01-21 18:37 UTC (permalink / raw)
To: Eric Sandeen; +Cc: linux-xfs
On Mon, Jan 20, 2020 at 08:29:17AM -0600, Eric Sandeen wrote:
> On 1/17/20 10:39 PM, Darrick J. Wong wrote:
> > On Fri, Jan 17, 2020 at 05:17:11PM -0600, Eric Sandeen wrote:
> >> The ->data_entry_p() op went away in v5.5 kernelspace, so rework
> >> xfs_repair to use ->data_entry_offset instead, in preparation
> >> for the v5.5 libxfs backport.
> >>
> >> This could later be cleaned up to use offsets as was done
> >> in kernel commit 8073af5153c for example.
> >
> > See, now that you've said that, I start wondering why not do that?
>
> Because this is the fast/safe path to getting the libxfs merge done IMHO ;)
>
> ...
>
>
> >> @@ -1834,7 +1834,7 @@ longform_dir2_entry_check_data(
> >> (dep->name[0] == '.' && dep->namelen == 1));
> >> add_inode_ref(current_irec, current_ino_offset);
> >> if (da_bno != 0 ||
> >> - dep != M_DIROPS(mp)->data_entry_p(d)) {
> >> + dep != (void *)d + M_DIROPS(mp)->data_entry_offset) {
> >
> > Er.... void pointer arithmetic?
>
> er, let me take another look at that.
fmeh, we apparently allow this gcc extension in the kernel so I guess
it's fine for xfsprogs :P
--D
> -eric
>
>
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] xfs_repair: stop using ->data_entry_p()
2020-01-21 18:37 ` Darrick J. Wong
@ 2020-01-22 0:41 ` Darrick J. Wong
0 siblings, 0 replies; 6+ messages in thread
From: Darrick J. Wong @ 2020-01-22 0:41 UTC (permalink / raw)
To: Eric Sandeen; +Cc: linux-xfs
On Tue, Jan 21, 2020 at 10:37:06AM -0800, Darrick J. Wong wrote:
> On Mon, Jan 20, 2020 at 08:29:17AM -0600, Eric Sandeen wrote:
> > On 1/17/20 10:39 PM, Darrick J. Wong wrote:
> > > On Fri, Jan 17, 2020 at 05:17:11PM -0600, Eric Sandeen wrote:
> > >> The ->data_entry_p() op went away in v5.5 kernelspace, so rework
> > >> xfs_repair to use ->data_entry_offset instead, in preparation
> > >> for the v5.5 libxfs backport.
> > >>
> > >> This could later be cleaned up to use offsets as was done
> > >> in kernel commit 8073af5153c for example.
> > >
> > > See, now that you've said that, I start wondering why not do that?
> >
> > Because this is the fast/safe path to getting the libxfs merge done IMHO ;)
> >
> > ...
> >
> >
> > >> @@ -1834,7 +1834,7 @@ longform_dir2_entry_check_data(
> > >> (dep->name[0] == '.' && dep->namelen == 1));
> > >> add_inode_ref(current_irec, current_ino_offset);
> > >> if (da_bno != 0 ||
> > >> - dep != M_DIROPS(mp)->data_entry_p(d)) {
> > >> + dep != (void *)d + M_DIROPS(mp)->data_entry_offset) {
> > >
> > > Er.... void pointer arithmetic?
> >
> > er, let me take another look at that.
>
> fmeh, we apparently allow this gcc extension in the kernel so I guess
> it's fine for xfsprogs :P
>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
--D
> --D
>
> > -eric
> >
> >
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] xfs_repair: stop using ->data_entry_p()
2020-01-17 23:17 [PATCH] xfs_repair: stop using ->data_entry_p() Eric Sandeen
2020-01-18 4:39 ` Darrick J. Wong
@ 2020-01-20 8:58 ` Christoph Hellwig
1 sibling, 0 replies; 6+ messages in thread
From: Christoph Hellwig @ 2020-01-20 8:58 UTC (permalink / raw)
To: Eric Sandeen; +Cc: linux-xfs, Darrick J. Wong
On Fri, Jan 17, 2020 at 05:17:11PM -0600, Eric Sandeen wrote:
> The ->data_entry_p() op went away in v5.5 kernelspace, so rework
> xfs_repair to use ->data_entry_offset instead, in preparation
> for the v5.5 libxfs backport.
>
> This could later be cleaned up to use offsets as was done
> in kernel commit 8073af5153c for example.
>
> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
> ---
>
> I'll munge this patch in mid-libxfs-sync, just before the
> ->data_entry_p removal patch.
Looks good, and I can give the cleanup a try once I find a little time.
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2020-01-22 0:41 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2020-01-17 23:17 [PATCH] xfs_repair: stop using ->data_entry_p() Eric Sandeen
2020-01-18 4:39 ` Darrick J. Wong
2020-01-20 14:29 ` Eric Sandeen
2020-01-21 18:37 ` Darrick J. Wong
2020-01-22 0:41 ` Darrick J. Wong
2020-01-20 8:58 ` Christoph Hellwig
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox