cifs: fix emitting cached direntries
Subsystems (2)
Steve, Tom, List
I update the patch to change rc to be bool and also expanded on the comments regarding ->pos as from Tom's comments.
PATCH 1/1 cifs: fix skipping to incorrect offset in emit_cached_dirents
When application has done lseek() to a different offset on a directory fd we skipped one entry too many before we start emitting directory entries from the cache.
We need to also make sure that when we are starting to emit directory entries from the cache, the ->pos sequence might have holes and skip some indices.
--- fs/cifs/readdir.c | 29 +++++++++++++++++++++++------ 1 file changed, 23 insertions(+), 6 deletions(-)
fs/cifs/readdir.c +23 -6
@@ -844,17 +844,34 @@ static bool emit_cached_dirents(struct cached_dirents *cde, struct dir_context *ctx) {struct cached_dirent *dirent; -int rc; +bool rc;list_for_each_entry (dirent, &cde->entries, entry) { -if (ctx->pos >= dirent->pos) +/* +* Skip all early entries prior to the current lseek() +* position. +*/ +if (ctx->pos > dirent->pos)continue ; +/* +* We recorded the current ->pos value for the dirent +* when we stored it in the cache. +* However, this sequence of ->pos values may have holes +* in it, for example dot-dirs returned from the server +* are suppressed. +* Handle this bu forcing ctx->pos to be the same as the +* ->pos of the current dirent we emit from the cache. +* This means that when we emit these entries from the cache +* we now emit them with the same ->pos value as in the +* initial scan. +*/ ctx->pos = dirent->pos; rc =dir_emit (ctx, dirent->name, dirent->namelen, dirent->fattr.cf_uniqueid, dirent->fattr.cf_dtype);if (!rc)return rc; + ctx->pos++; }return true; }
@@ -1202,10 +1219,10 @@ int cifs_readdir(struct file *file, struct dir_context *ctx) ctx->pos, tmp_buf);cifs_save_resume_key (current_entry, cifsFile);break ; - }else - current_entry = -nxt_dir_entry (current_entry, end_of_smb, - cifsFile->srch_inf.info_level); + } + current_entry = +nxt_dir_entry (current_entry, end_of_smb, + cifsFile->srch_inf.info_level); }kfree (tmp_buf);
Signature · 1 line
2.35.3
Re: [PATCH] cifs: fix skipping to incorrect offset in emit_cached_dirents
Quoted 34 lines · Ronnie Sahlberg
When application has done
lseek()to a different offset on a directory fd we skipped one entry too many before we start emitting directory entries from the cache.We need to also make sure that when we are starting to emit directory entries from the cache, the ->pos sequence might have holes and skip some indices.
Signed-off-by Ronnie Sahlberg <[email protected]>--- fs/cifs/readdir.c | 29 +++++++++++++++++++++++------ 1 file changed, 23 insertions(+), 6 deletions(-)fs/cifs/readdir.c +8
-2@@ -844,17 +844,34 @@ static bool emit_cached_dirents(struct cached_dirents *cde, struct dir_context *ctx) {struct cached_dirent *dirent; -int rc; +bool rc;list_for_each_entry (dirent, &cde->entries, entry) { -if (ctx->pos >= dirent->pos) +/* +* Skip all early entries prior to the current lseek() +* position. +*/ +if (ctx->pos > dirent->pos)continue ; +/* +* We recorded the current ->pos value for the dirent
fs/cifs/readdir.c static bool emit_cached_dirents(struct cached_dirents *cde, · line 867
Quoted 10 lines
fs/cifs/readdir.c +10
-0+* when we stored it in the cache. +* However, this sequence of ->pos values may have holes +* in it, for example dot-dirs returned from the server +* are suppressed. +* Handle this bu forcing ctx->pos to be the same as the +* ->pos of the current dirent we emit from the cache. +* This means that when we emit these entries from the cache +* we now emit them with the same ->pos value as in the +* initial scan. +*/
It's a little wordy, but it's better, so ok.
fs/cifs/readdir.c static bool emit_cached_dirents(struct cached_dirents *cde, · line 873
Quoted 6 lines
fs/cifs/readdir.c +0
-0ctx->pos = dirent->pos; rc =dir_emit (ctx, dirent->name, dirent->namelen, dirent->fattr.cf_uniqueid, dirent->fattr.cf_dtype);if (!rc)return rc;
Well, I still think it would be clearer as "return false". And the name "rc" really seems odd now for a bool. But again, ok I guess.
Either way, I learned a thing or two chasing down dir_emit() and the overall approach seems good.
Quoted 6 lines
fs/cifs/readdir.c +1
-0+ ctx->pos++; }return true; }@@ -1202,10 +1219,10 @@ int cifs_readdir(struct file *file, struct dir_context *ctx) ctx->pos, tmp_buf);
fs/cifs/readdir.c int cifs_readdir(struct file *file, struct dir_context *ctx) · line 1225
Quoted 10 lines
fs/cifs/readdir.c +4
-4cifs_save_resume_key (current_entry, cifsFile);break ; - }else - current_entry = -nxt_dir_entry (current_entry, end_of_smb, - cifsFile->srch_inf.info_level); + } + current_entry = +nxt_dir_entry (current_entry, end_of_smb, + cifsFile->srch_inf.info_level);
This change isn't actually changing anything. Would it be better to leave out on principle?
fs/cifs/readdir.c int cifs_readdir(struct file *file, struct dir_context *ctx) · line 1228
Quoted 3 lines
fs/cifs/readdir.c +0
-0}kfree (tmp_buf);
Overall,
Re: [PATCH] cifs: fix skipping to incorrect offset in emit_cached_dirents
Quoted 73 lines · Tom Talpey
Quoted 34 lines · Ronnie Sahlberg
When application has done
lseek()to a different offset on a directory fd we skipped one entry too many before we start emitting directory entries from the cache.We need to also make sure that when we are starting to emit directory entries from the cache, the ->pos sequence might have holes and skip some indices.
Signed-off-by Ronnie Sahlberg <[email protected]>--- fs/cifs/readdir.c | 29 +++++++++++++++++++++++------ 1 file changed, 23 insertions(+), 6 deletions(-)fs/cifs/readdir.c +8
-2@@ -844,17 +844,34 @@ static bool emit_cached_dirents(struct cached_dirents *cde, struct dir_context *ctx) {struct cached_dirent *dirent; -int rc; +bool rc;list_for_each_entry (dirent, &cde->entries, entry) { -if (ctx->pos >= dirent->pos) +/* +* Skip all early entries prior to the current lseek() +* position. +*/ +if (ctx->pos > dirent->pos)continue ; +/* +* We recorded the current ->pos value for the dirent fs/cifs/readdir.c static bool emit_cached_dirents(struct cached_dirents *cde, · line 867
Quoted 10 lines
fs/cifs/readdir.c +10
-0+* when we stored it in the cache. +* However, this sequence of ->pos values may have holes +* in it, for example dot-dirs returned from the server +* are suppressed. +* Handle this bu forcing ctx->pos to be the same as the +* ->pos of the current dirent we emit from the cache. +* This means that when we emit these entries from the cache +* we now emit them with the same ->pos value as in the +* initial scan. +*/ It's a little wordy, but it's better, so ok.
fs/cifs/readdir.c static bool emit_cached_dirents(struct cached_dirents *cde, · line 873
Quoted 6 lines
fs/cifs/readdir.c +0
-0ctx->pos = dirent->pos; rc =dir_emit (ctx, dirent->name, dirent->namelen, dirent->fattr.cf_uniqueid, dirent->fattr.cf_dtype);if (!rc)return rc;Well, I still think it would be clearer as "return false". And the name "rc" really seems odd now for a bool. But again, ok I guess.
Either way, I learned a thing or two chasing down
dir_emit()and the overall approach seems good.Quoted 11 lines
fs/cifs/readdir.c +1
-3+ ctx->pos++; }return true; }@@ -1202,10 +1219,10 @@ int cifs_readdir(struct file *file, struct dir_context *ctx) ctx->pos, tmp_buf);cifs_save_resume_key (current_entry, cifsFile);break ; - }else - current_entry = - nxt_dir_entry(current_entry, end_of_smb,
Quoted 8 lines
fs/cifs/readdir.c int cifs_readdir(struct file *file, struct dir_context *ctx) · line 1225
Quoted 5 lines
fs/cifs/readdir.c +4
-1- cifsFile->srch_inf.info_level); + } + current_entry = +nxt_dir_entry (current_entry, end_of_smb, + cifsFile->srch_inf.info_level);This change isn't actually changing anything. Would it be better to leave out on principle?
I wanted this change since previously it was a little confusing. A quick glance might raise questions "why only do this conditionally?" and then when looking more carefully "oh, we actually do this unconditionally before every new iteration".
It is as you say a no-op change.
It should arguably be done in a separate patch but it is so small I just took the lazy path and put it in this patch instead of a separate patch.
Thanks for the review.
Quoted 9 lines
fs/cifs/readdir.c int cifs_readdir(struct file *file, struct dir_context *ctx) · line 1228
Quoted 5 lines
fs/cifs/readdir.c +0
-0}kfree (tmp_buf);Overall,
Reviewed-by Tom Talpey <[email protected]>