convert write_cache_pages() to an iterator v6
Subsystems (3)
Hi all,
this is an evolution of the series Matthew Wilcox originally sent in June 2023, which has changed quite a bit since and now has a while based iterator.
Note that in this version two patches are so different from the previous version that I've not kept any Reviews or Acks for them, even if the final result look almost the same as the previous patches with the incremental patch on the list.
Changes since v5:
- completely reshuffle the series to directly prepare for the writeback_iter() style.
- don't require *error to be initialized on first call
- improve various comments
- fix a bisection hazard where write_cache_pages don't return delayed error for a few commits
- fix a whitespace error
- drop the iomap patch again for now as the iomap map multiple blocks series isn't in mainline yet
Changes since v4:
- added back the (rebased) iomap conversion now that the conflict is in mainline
- add a new patch to change the iterator
Changes since v3:
- various commit log spelling fixes
- remove a statement from a commit log that isn't true any more with the changes in v3
- rename a function
- merge two helpers
Diffstat:
fs/iomap/buffered-io.c | 10 - include/linux/pagevec.h | 18 ++ include/linux/writeback.h | 12 + mm/page-writeback.c | 344 ++++++++++++++++++++++++++-------------------- 4 files changed, 231 insertions(+), 153 deletions(-)
PATCH 1/13 writeback: remove a duplicate prototype for tag_pages_for_writeback
From: "Matthew Wilcox (Oracle)" <[email protected]>
--- include/linux/writeback.h | 2 -- 1 file changed, 2 deletions(-)
include/linux/writeback.h +0 -2
@@ -363,8 +363,6 @@ bool wb_over_bg_thresh(struct bdi_writeback *wb);typedef int (*writepage_t )(struct folio *folio,struct writeback_control *wbc,void *data); -void tag_pages_for_writeback (struct address_space *mapping, -pgoff_t start,pgoff_t end);int write_cache_pages (struct address_space *mapping,struct writeback_control *wbc,writepage_t writepage,void *data);
Signature · 1 line
2.39.2
PATCH 2/13 writeback: fix done_index when hitting the wbc->nr_to_write
In reply to Christoph HellwigWhen write_cache_pages finishes writing out a folio, it fails to update done_index to account for the number of pages in the folio just written. That means when range_cyclic writeback is restarted, it will be restarted at this folio instead of after it as it should. Fix that by updating done_index before breaking out of the loop.
--- mm/page-writeback.c | 1 + 1 file changed, 1 insertion(+)
mm/page-writeback.c +1 -0
@@ -2505,6 +2505,7 @@ int write_cache_pages(struct address_space *mapping,* keep going until we have written all the pages * we tagged for writeback prior to entering this loop. */ + done_index = folio->index + nr; wbc->nr_to_write -= nr;if (wbc->nr_to_write <=0 && wbc->sync_mode ==WB_SYNC_NONE ) {
Signature · 1 line
2.39.2
PATCH 3/13 writeback: also update wbc->nr_to_write on writeback failure
In reply to Christoph HellwigWhen exiting write_cache_pages early due to a non-integrity write failure, wbc->nr_to_write currently doesn't account for the folio we just failed to write. This doesn't matter because the callers always ingore the value on a failure, but moving the update to common code will allow to simplify the code, so do it.
--- mm/page-writeback.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
mm/page-writeback.c +1 -1
@@ -2473,6 +2473,7 @@ int write_cache_pages(struct address_space *mapping,trace_wbc_writepage (wbc,inode_to_bdi (mapping->host)); error =writepage (folio, wbc, data); nr =folio_nr_pages (folio); + wbc->nr_to_write -= nr;if (unlikely (error)) {/* * Handle errors according to the type of
@@ -2506,7 +2507,6 @@ int write_cache_pages(struct address_space *mapping,* we tagged for writeback prior to entering this loop. */ done_index = folio->index + nr; - wbc->nr_to_write -= nr;if (wbc->nr_to_write <=0 && wbc->sync_mode ==WB_SYNC_NONE ) { done =1 ;
Signature · 1 line
2.39.2
PATCH 4/13 writeback: only update ->writeback_index for range_cyclic writeback
In reply to Christoph Hellwigmapping->writeback_index is only [1] used as the starting point for range_cyclic writeback, so there is no point in updating it for other types of writeback.
[1] except for btrfs_defrag_file which does really odd things with mapping->writeback_index. But btrfs doesn't use write_cache_pages at all, so this isn't relevant here.
--- mm/page-writeback.c | 24 ++++++++++++++---------- 1 file changed, 14 insertions(+), 10 deletions(-)
mm/page-writeback.c +14 -10
@@ -2403,7 +2403,6 @@ int write_cache_pages(struct address_space *mapping,pgoff_t index;pgoff_t end;/* Inclusive */ pgoff_t done_index; -int range_whole =0 ;xa_mark_t tag;folio_batch_init (&fbatch);
@@ -2413,8 +2412,6 @@ int write_cache_pages(struct address_space *mapping, }else { index = wbc->range_start >>PAGE_SHIFT ; end = wbc->range_end >>PAGE_SHIFT ; -if (wbc->range_start ==0 && wbc->range_end ==LLONG_MAX ) - range_whole =1 ; }if (wbc->sync_mode ==WB_SYNC_ALL || wbc->tagged_writepages) {tag_pages_for_writeback (mapping, index, end);
@@ -2518,14 +2515,21 @@ int write_cache_pages(struct address_space *mapping, }/* -* If we hit the last page and there is more work to be done: wrap -* back the index back to the start of the file for the next -* time we are called. +* For range cyclic writeback we need to remember where we stopped so +* that we can continue there next time we are called. If we hit the +* last page and there is more work to be done, wrap back to the start +* of the file. +* +* For non-cyclic writeback we always start looking up at the beginning +* of the file if we are called again, which can only happen due to +* -ENOMEM from the file system. */ -if (wbc->range_cyclic && !done) - done_index =0 ; -if (wbc->range_cyclic || (range_whole && wbc->nr_to_write >0 )) - mapping->writeback_index = done_index; +if (wbc->range_cyclic) { +if (done) + mapping->writeback_index = done_index; +else + mapping->writeback_index =0 ; + }return ret; }
Signature · 1 line
2.39.2
PATCH 5/13 writeback: rework the loop termination condition in write_cache_pages
In reply to Christoph HellwigRework the way we deal with the cleanup after the writepage call.
First handle the magic AOP_WRITEPAGE_ACTIVATE separately from real error returns to get it out of the way of the actual error handling path.
The split the handling on intgrity vs non-integrity branches first, and return early using a goto for the non-ingegrity early loop condition to remove the need for the done and done_index local variables, and for assigning the error to ret when we can just return error directly.
--- mm/page-writeback.c | 84 ++++++++++++++++++--------------------------- 1 file changed, 33 insertions(+), 51 deletions(-)
mm/page-writeback.c +33 -51
@@ -2396,13 +2396,12 @@ int write_cache_pages(struct address_space *mapping, void *data) {int ret =0 ; -int done =0 ;int error;struct folio_batch fbatch; +struct folio *folio;int nr_folios;pgoff_t index;pgoff_t end;/* Inclusive */ -pgoff_t done_index;xa_mark_t tag;folio_batch_init (&fbatch);
@@ -2419,8 +2418,7 @@ int write_cache_pages(struct address_space *mapping, }else { tag =PAGECACHE_TAG_DIRTY ; } - done_index = index; -while (!done && (index <= end)) { +while (index <= end) {int i; nr_folios = filemap_get_folios_tag(mapping, &index, end,
@@ -2430,11 +2428,7 @@ int write_cache_pages(struct address_space *mapping,break ;for (i =0 ; i < nr_folios; i++) { -struct folio *folio = fbatch.folios[i]; -unsigned long nr; - - done_index = folio->index; - + folio = fbatch.folios[i];folio_lock (folio);/*
@@ -2469,45 +2463,32 @@ int write_cache_pages(struct address_space *mapping,trace_wbc_writepage (wbc,inode_to_bdi (mapping->host)); error =writepage (folio, wbc, data); - nr =folio_nr_pages (folio); - wbc->nr_to_write -= nr; -if (unlikely (error)) { -/* -* Handle errors according to the type of -* writeback. There's no need to continue for -* background writeback. Just push done_index -* past this page so media errors won't choke -* writeout for the entire file. For integrity -* writeback, we must process the entire dirty -* set regardless of errors because the fs may -* still have state to clear for each page. In -* that case we continue processing and return -* the first error. -*/ -if (error ==AOP_WRITEPAGE_ACTIVATE ) { -folio_unlock (folio); - error =0 ; - }else if (wbc->sync_mode !=WB_SYNC_ALL ) { - ret = error; - done_index = folio->index + nr; - done =1 ; -break ; - } -if (!ret) - ret = error; + wbc->nr_to_write -=folio_nr_pages (folio); + +if (error ==AOP_WRITEPAGE_ACTIVATE ) { +folio_unlock (folio); + error =0 ; }/* -* We stop writing back only if we are not doing -* integrity sync. In case of integrity sync we have to -* keep going until we have written all the pages -* we tagged for writeback prior to entering this loop. +* For integrity writeback we have to keep going until +* we have written all the folios we tagged for +* writeback above, even if we run past wbc->nr_to_write +* or encounter errors. +* We stash away the first error we encounter in +* wbc->saved_err so that it can be retrieved when we're +* done. This is because the file system may still have +* state to clear for each folio. +* +* For background writeback we exit as soon as we run +* past wbc->nr_to_write or encounter the first error. */ - done_index = folio->index + nr; -if (wbc->nr_to_write <=0 && - wbc->sync_mode ==WB_SYNC_NONE ) { - done =1 ; -break ; +if (wbc->sync_mode ==WB_SYNC_ALL ) { +if (error && !ret) + ret = error; + }else { +if (error || wbc->nr_to_write <=0 ) + goto done; } }folio_batch_release (&fbatch);
@@ -2524,14 +2505,15 @@ int write_cache_pages(struct address_space *mapping,* of the file if we are called again, which can only happen due to * -ENOMEM from the file system. */ -if (wbc->range_cyclic) { -if (done) - mapping->writeback_index = done_index; -else - mapping->writeback_index =0 ; - } - +if (wbc->range_cyclic) + mapping->writeback_index =0 ;return ret; + +done: +folio_batch_release (&fbatch); +if (wbc->range_cyclic) + mapping->writeback_index = folio->index +folio_nr_pages (folio); +return error; }EXPORT_SYMBOL (write_cache_pages);
Signature · 1 line
2.39.2
Re: [PATCH 05/13] writeback: rework the loop termination condition in write_cache_pages
Quoted 11 lines · Christoph Hellwig
Rework the way we deal with the cleanup after the writepage call.
First handle the magic
AOP_WRITEPAGE_ACTIVATEseparately from real error returns to get it out of the way of the actual error handling path.The split the handling on intgrity vs non-integrity branches first, and return early using a goto for the non-ingegrity early loop condition to remove the need for the done and
done_indexlocal variables, and for assigning the error to ret when we can just return error directly.Signed-off-by Christoph Hellwig <[email protected]>
Looks good. Feel free to add:
Honza
Quoted 140 lines
--- mm/page-writeback.c | 84 ++++++++++++++++++--------------------------- 1 file changed, 33 insertions(+), 51 deletions(-)mm/page-writeback.c +33
-51@@ -2396,13 +2396,12 @@ int write_cache_pages(struct address_space *mapping, void *data) {int ret =0 ; -int done =0 ;int error;struct folio_batch fbatch; +struct folio *folio;int nr_folios;pgoff_t index;pgoff_t end;/* Inclusive */ -pgoff_t done_index;xa_mark_t tag;folio_batch_init (&fbatch);@@ -2419,8 +2418,7 @@ int write_cache_pages(struct address_space *mapping, }else { tag =PAGECACHE_TAG_DIRTY ; } - done_index = index; -while (!done && (index <= end)) { +while (index <= end) {int i; nr_folios = filemap_get_folios_tag(mapping, &index, end,@@ -2430,11 +2428,7 @@ int write_cache_pages(struct address_space *mapping,break ;for (i =0 ; i < nr_folios; i++) { -struct folio *folio = fbatch.folios[i]; -unsigned long nr; - - done_index = folio->index; - + folio = fbatch.folios[i];folio_lock (folio);/* @@ -2469,45 +2463,32 @@ int write_cache_pages(struct address_space *mapping,trace_wbc_writepage (wbc,inode_to_bdi (mapping->host)); error =writepage (folio, wbc, data); - nr =folio_nr_pages (folio); - wbc->nr_to_write -= nr; -if (unlikely (error)) { -/* -* Handle errors according to the type of -* writeback. There's no need to continue for -* background writeback. Just push done_index -* past this page so media errors won't choke -* writeout for the entire file. For integrity -* writeback, we must process the entire dirty -* set regardless of errors because the fs may -* still have state to clear for each page. In -* that case we continue processing and return -* the first error. -*/ -if (error ==AOP_WRITEPAGE_ACTIVATE ) { -folio_unlock (folio); - error =0 ; - }else if (wbc->sync_mode !=WB_SYNC_ALL ) { - ret = error; - done_index = folio->index + nr; - done =1 ; -break ; - } -if (!ret) - ret = error; + wbc->nr_to_write -=folio_nr_pages (folio); + +if (error ==AOP_WRITEPAGE_ACTIVATE ) { +folio_unlock (folio); + error =0 ; }/* -* We stop writing back only if we are not doing -* integrity sync. In case of integrity sync we have to -* keep going until we have written all the pages -* we tagged for writeback prior to entering this loop. +* For integrity writeback we have to keep going until +* we have written all the folios we tagged for +* writeback above, even if we run past wbc->nr_to_write +* or encounter errors. +* We stash away the first error we encounter in +* wbc->saved_err so that it can be retrieved when we're +* done. This is because the file system may still have +* state to clear for each folio. +* +* For background writeback we exit as soon as we run +* past wbc->nr_to_write or encounter the first error. */ - done_index = folio->index + nr; -if (wbc->nr_to_write <=0 && - wbc->sync_mode ==WB_SYNC_NONE ) { - done =1 ; -break ; +if (wbc->sync_mode ==WB_SYNC_ALL ) { +if (error && !ret) + ret = error; + }else { +if (error || wbc->nr_to_write <=0 ) + goto done; } }folio_batch_release (&fbatch);@@ -2524,14 +2505,15 @@ int write_cache_pages(struct address_space *mapping,* of the file if we are called again, which can only happen due to * -ENOMEM from the file system. */ -if (wbc->range_cyclic) { -if (done) - mapping->writeback_index = done_index; -else - mapping->writeback_index =0 ; - } - +if (wbc->range_cyclic) + mapping->writeback_index =0 ;return ret; + +done: +folio_batch_release (&fbatch); +if (wbc->range_cyclic) + mapping->writeback_index = folio->index +folio_nr_pages (folio); +return error; }EXPORT_SYMBOL (write_cache_pages);Signature · 1 line
2.39.2
Signature · 2 lines
Jan Kara <[email protected]> SUSE Labs, CR
Re: [PATCH 05/13] writeback: rework the loop termination condition in write_cache_pages
In reply to Christoph HellwigQuoted 19 lines · Christoph Hellwig
Rework the way we deal with the cleanup after the writepage call.
First handle the magic
AOP_WRITEPAGE_ACTIVATEseparately from real error returns to get it out of the way of the actual error handling path.The split the handling on intgrity vs non-integrity branches first, and return early using a goto for the non-ingegrity early loop condition to remove the need for the done and
done_indexlocal variables, and for assigning the error to ret when we can just return error directly.Signed-off-by Christoph Hellwig <[email protected]>--- mm/page-writeback.c | 84 ++++++++++++++++++--------------------------- 1 file changed, 33 insertions(+), 51 deletions(-)mm/page-writeback.c +0
-0
...
Quoted 9 lines
mm/page-writeback.c +0
-5@@ -2524,14 +2505,15 @@ int write_cache_pages(struct address_space *mapping,* of the file if we are called again, which can only happen due to * -ENOMEM from the file system. */ -if (wbc->range_cyclic) { -if (done) - mapping->writeback_index = done_index; -else - mapping->writeback_index =0 ;
mm/page-writeback.c int write_cache_pages(struct address_space *mapping, · line 2515
Quoted 10 lines
mm/page-writeback.c +7
-2- } - +if (wbc->range_cyclic) + mapping->writeback_index =0 ;return ret; + +done: +folio_batch_release (&fbatch); +if (wbc->range_cyclic) + mapping->writeback_index = folio->index +folio_nr_pages (folio);
Shouldn't this release the batch after we're done accessing folio? With that addressed:
Quoted 8 lines
mm/page-writeback.c +1
-0+return error; }EXPORT_SYMBOL (write_cache_pages);Signature · 1 line
2.39.2
PATCH 6/13 writeback: Factor folio_prepare_writeback() out of write_cache_pages()
In reply to Christoph HellwigFrom: "Matthew Wilcox (Oracle)" <[email protected]>
Reduce write_cache_pages() by about 30 lines; much of it is commentary, but it all bundles nicely into an obvious function.
[hch: rename should_writeback_folio to folio_prepare_writeback based on
a comment from Jan Kara]
--- mm/page-writeback.c | 61 +++++++++++++++++++++++++-------------------- 1 file changed, 34 insertions(+), 27 deletions(-)
mm/page-writeback.c +34 -27
@@ -2360,6 +2360,38 @@ void tag_pages_for_writeback(struct address_space *mapping, }EXPORT_SYMBOL (tag_pages_for_writeback); +static bool folio_prepare_writeback (struct address_space *mapping, +struct writeback_control *wbc,struct folio *folio) +{ +/* +* Folio truncated or invalidated. We can freely skip it then, +* even for data integrity operations: the folio has disappeared +* concurrently, so there could be no real expectation of this +* data integrity operation even if there is now a new, dirty +* folio at the same pagecache index. +*/ +if (unlikely (folio->mapping != mapping)) +return false; + +/* +* Did somebody else write it for us? +*/ +if (!folio_test_dirty (folio)) +return false; + +if (folio_test_writeback (folio)) { +if (wbc->sync_mode ==WB_SYNC_NONE ) +return false; +folio_wait_writeback (folio); + } +BUG_ON (folio_test_writeback (folio)); + +if (!folio_clear_dirty_for_io (folio)) +return false; + +return true; +} +/** * write_cache_pages - walk the list of dirty pages of the given address space and write all of them. * @mapping: address space structure to write
@@ -2430,38 +2462,13 @@ int write_cache_pages(struct address_space *mapping,for (i =0 ; i < nr_folios; i++) { folio = fbatch.folios[i];folio_lock (folio); - -/* -* Page truncated or invalidated. We can freely skip it -* then, even for data integrity operations: the page -* has disappeared concurrently, so there could be no -* real expectation of this data integrity operation -* even if there is now a new, dirty page at the same -* pagecache address. -*/ -if (unlikely (folio->mapping != mapping)) { -continue_unlock: +if (!folio_prepare_writeback (mapping, wbc, folio)) {folio_unlock (folio);continue ; } -if (!folio_test_dirty (folio)) { -/* someone wrote it for us */ - goto continue_unlock; - } - -if (folio_test_writeback (folio)) { -if (wbc->sync_mode !=WB_SYNC_NONE ) -folio_wait_writeback (folio); -else - goto continue_unlock; - } - -BUG_ON (folio_test_writeback (folio)); -if (!folio_clear_dirty_for_io (folio)) - goto continue_unlock; -trace_wbc_writepage (wbc,inode_to_bdi (mapping->host)); + error =writepage (folio, wbc, data); wbc->nr_to_write -=folio_nr_pages (folio);
Signature · 1 line
2.39.2
PATCH 7/13 writeback: Factor writeback_get_batch() out of write_cache_pages()
In reply to Christoph HellwigFrom: "Matthew Wilcox (Oracle)" <[email protected]>
This simple helper will be the basis of the writeback iterator. To make this work, we need to remember the current index and end positions in writeback_control.
[hch: heavily rebased, add helpers to get the tag and end index,
don't keep the end index in struct writeback_control]
--- include/linux/writeback.h | 6 ++++ mm/page-writeback.c | 60 +++++++++++++++++++++++++-------------- 2 files changed, 44 insertions(+), 22 deletions(-)
include/linux/writeback.h +6 -0
@@ -11,6 +11,7 @@#include <linux/flex_proportions.h> #include <linux/backing-dev-defs.h> #include <linux/blk_types.h> +#include <linux/pagevec.h> struct bio ;
@@ -40,6 +41,7 @@ enum writeback_sync_modes {* in a manner such that unspecified fields are set to zero. */ struct writeback_control { +/* public fields that can be set and/or consumed by the caller: */ long nr_to_write;/* Write this many pages, and decrement this for each page written */ long pages_skipped;/* Pages which were not written */
@@ -77,6 +79,10 @@ struct writeback_control {*/ struct swap_iocb **swap_plug; +/* internal fields used by the ->writepages implementation: */ +struct folio_batch fbatch; +pgoff_t index; +#ifdef CONFIG_CGROUP_WRITEBACK struct bdi_writeback *wb;/* wb this writeback is issued under */ struct inode *inode;/* inode being written out */
mm/page-writeback.c +38 -22
102 lines not on this page · raw message
Signature · 1 line
2.39.2
PATCH 8/13 writeback: Simplify the loops in write_cache_pages()
In reply to Christoph HellwigFrom: "Matthew Wilcox (Oracle)" <[email protected]>
Collapse the two nested loops into one. This is needed as a step towards turning this into an iterator.
Note that this drops the "index <= end" check in the previous outer loop and just relies on filemap_get_folios_tag() to return 0 entries when index > end. This actually has a subtle implication when end == -1 because then the returned index will be -1 as well and thus if there is page present on index -1, we could be looping indefinitely. But as the comment in filemap_get_folios_tag documents this as already broken anyway we should not worry about it here either. The fix for that would probably a change to the filemap_get_folios_tag() calling convention.
--- mm/page-writeback.c | 75 ++++++++++++++++++++++----------------------- 1 file changed, 36 insertions(+), 39 deletions(-)
mm/page-writeback.c +36 -39
@@ -2454,6 +2454,7 @@ int write_cache_pages(struct address_space *mapping,int error;struct folio *folio;pgoff_t end;/* Inclusive */ +int i =0 ;if (wbc->range_cyclic) { wbc->index = mapping->writeback_index;/* prev offset */
@@ -2467,53 +2468,49 @@ int write_cache_pages(struct address_space *mapping,folio_batch_init (&wbc->fbatch); -while (wbc->index <= end) { -int i; - -writeback_get_batch (mapping, wbc); - +for (;;) { +if (i == wbc->fbatch.nr) { +writeback_get_batch (mapping, wbc); + i =0 ; + }if (wbc->fbatch.nr ==0 )break ; -for (i =0 ; i < wbc->fbatch.nr; i++) { - folio = wbc->fbatch.folios[i]; + folio = wbc->fbatch.folios[i++]; -folio_lock (folio); -if (!folio_prepare_writeback (mapping, wbc, folio)) { -folio_unlock (folio); -continue ; - } +folio_lock (folio); +if (!folio_prepare_writeback (mapping, wbc, folio)) { +folio_unlock (folio); +continue ; + } -trace_wbc_writepage (wbc,inode_to_bdi (mapping->host)); +trace_wbc_writepage (wbc,inode_to_bdi (mapping->host)); - error =writepage (folio, wbc, data); - wbc->nr_to_write -=folio_nr_pages (folio); + error =writepage (folio, wbc, data); + wbc->nr_to_write -=folio_nr_pages (folio); -if (error ==AOP_WRITEPAGE_ACTIVATE ) { -folio_unlock (folio); - error =0 ; - } +if (error ==AOP_WRITEPAGE_ACTIVATE ) { +folio_unlock (folio); + error =0 ; + } -/* -* For integrity writeback we have to keep going until -* we have written all the folios we tagged for -* writeback above, even if we run past wbc->nr_to_write -* or encounter errors. -* We stash away the first error we encounter in -* wbc->saved_err so that it can be retrieved when we're -* done. This is because the file system may still have -* state to clear for each folio. -* -* For background writeback we exit as soon as we run -* past wbc->nr_to_write or encounter the first error. -*/ -if (wbc->sync_mode ==WB_SYNC_ALL ) { -if (error && !ret) - ret = error; - }else { -if (error || wbc->nr_to_write <=0 ) - goto done; - } +/* +* For integrity writeback we have to keep going until we have +* written all the folios we tagged for writeback above, even if +* we run past wbc->nr_to_write or encounter errors. +* We stash away the first error we encounter in wbc->saved_err +* so that it can be retrieved when we're done. This is because +* the file system may still have state to clear for each folio. +* +* For background writeback we exit as soon as we run past +* wbc->nr_to_write or encounter the first error. +*/ +if (wbc->sync_mode ==WB_SYNC_ALL ) { +if (error && !ret) + ret = error; + }else { +if (error || wbc->nr_to_write <=0 ) + goto done; } }
Signature · 1 line
2.39.2
PATCH 9/13 pagevec: Add ability to iterate a queue
In reply to Christoph HellwigFrom: "Matthew Wilcox (Oracle)" <[email protected]>
Add a loop counter inside the folio_batch to let us iterate from 0-nr instead of decrementing nr and treating the batch as a stack. It would generate some very weird and suboptimal I/O patterns for page writeback to iterate over the batch as a stack.
--- include/linux/pagevec.h | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+)
include/linux/pagevec.h +18 -0
@@ -27,6 +27,7 @@ struct folio;*/ struct folio_batch {unsigned char nr; +unsigned char i;bool percpu_pvec_drained;struct folio *folios[PAGEVEC_SIZE ]; };
@@ -40,12 +41,14 @@ struct folio_batch {static inline void folio_batch_init (struct folio_batch *fbatch) { fbatch->nr =0 ; + fbatch->i =0 ; fbatch->percpu_pvec_drained = false; }static inline void folio_batch_reinit (struct folio_batch *fbatch) { fbatch->nr =0 ; + fbatch->i =0 ; }static inline unsigned int folio_batch_count (struct folio_batch *fbatch)
@@ -75,6 +78,21 @@ static inline unsigned folio_batch_add(struct folio_batch *fbatch,return folio_batch_space (fbatch); } +/** +* folio_batch_next - Return the next folio to process. +* @fbatch: The folio batch being processed. +* +* Use this function to implement a queue of folios. +* +* Return: The next folio in the queue, or NULL if the queue is empty. +*/ +static inline struct folio *folio_batch_next (struct folio_batch *fbatch) +{ +if (fbatch->i == fbatch->nr) +return NULL ; +return fbatch->folios[fbatch->i++]; +} +void __folio_batch_release (struct folio_batch *pvec);static inline void folio_batch_release (struct folio_batch *fbatch)
Signature · 1 line
2.39.2
PATCH 10/13 writeback: Use the folio_batch queue iterator
In reply to Christoph HellwigFrom: "Matthew Wilcox (Oracle)" <[email protected]>
Instead of keeping our own local iterator variable, use the one just added to folio_batch.
--- mm/page-writeback.c | 28 +++++++++++++++------------- 1 file changed, 15 insertions(+), 13 deletions(-)
mm/page-writeback.c +15 -13
@@ -2406,13 +2406,21 @@ static pgoff_t wbc_end(struct writeback_control *wbc)return wbc->range_end >>PAGE_SHIFT ; } -static void writeback_get_batch (struct address_space *mapping, +static struct folio *writeback_get_folio (struct address_space *mapping,struct writeback_control *wbc) { -folio_batch_release (&wbc->fbatch); -cond_resched (); -filemap_get_folios_tag (mapping, &wbc->index,wbc_end (wbc), -wbc_to_tag (wbc), &wbc->fbatch); +struct folio *folio; + + folio =folio_batch_next (&wbc->fbatch); +if (!folio) { +folio_batch_release (&wbc->fbatch); +cond_resched (); +filemap_get_folios_tag (mapping, &wbc->index,wbc_end (wbc), +wbc_to_tag (wbc), &wbc->fbatch); + folio =folio_batch_next (&wbc->fbatch); + } + +return folio; }/**
@@ -2454,7 +2462,6 @@ int write_cache_pages(struct address_space *mapping,int error;struct folio *folio;pgoff_t end;/* Inclusive */ -int i =0 ;if (wbc->range_cyclic) { wbc->index = mapping->writeback_index;/* prev offset */
@@ -2469,15 +2476,10 @@ int write_cache_pages(struct address_space *mapping,folio_batch_init (&wbc->fbatch);for (;;) { -if (i == wbc->fbatch.nr) { -writeback_get_batch (mapping, wbc); - i =0 ; - } -if (wbc->fbatch.nr ==0 ) + folio =writeback_get_folio (mapping, wbc); +if (!folio)break ; - folio = wbc->fbatch.folios[i++]; -folio_lock (folio);if (!folio_prepare_writeback (mapping, wbc, folio)) {folio_unlock (folio);
Signature · 1 line
2.39.2
PATCH 11/13 writeback: Move the folio_prepare_writeback loop out of write_cache_pages()
In reply to Christoph HellwigFrom: "Matthew Wilcox (Oracle)" <[email protected]>
Move the loop for should-we-write-this-folio to writeback_get_folio.
[hch: folded the loop into the existing helper instead of a separate one
as suggested by Jan Kara]
--- mm/page-writeback.c | 18 ++++++++++-------- 1 file changed, 10 insertions(+), 8 deletions(-)
mm/page-writeback.c +10 -8
@@ -2411,6 +2411,7 @@ static struct folio *writeback_get_folio(struct address_space *mapping, {struct folio *folio; +retry: folio =folio_batch_next (&wbc->fbatch);if (!folio) {folio_batch_release (&wbc->fbatch);
@@ -2418,8 +2419,17 @@ static struct folio *writeback_get_folio(struct address_space *mapping,filemap_get_folios_tag (mapping, &wbc->index,wbc_end (wbc),wbc_to_tag (wbc), &wbc->fbatch); folio =folio_batch_next (&wbc->fbatch); +if (!folio) +return NULL ; + } + +folio_lock (folio); +if (unlikely (!folio_prepare_writeback (mapping, wbc, folio))) { +folio_unlock (folio); + goto retry; } +trace_wbc_writepage (wbc,inode_to_bdi (mapping->host));return folio; }
@@ -2480,14 +2490,6 @@ int write_cache_pages(struct address_space *mapping,if (!folio)break ; -folio_lock (folio); -if (!folio_prepare_writeback (mapping, wbc, folio)) { -folio_unlock (folio); -continue ; - } - -trace_wbc_writepage (wbc,inode_to_bdi (mapping->host)); - error =writepage (folio, wbc, data); wbc->nr_to_write -=folio_nr_pages (folio);
Signature · 1 line
2.39.2
PATCH 12/13 writeback: add a writeback iterator
In reply to Christoph HellwigRefactor the code left in write_cache_pages into an iterator that the file system can call to get the next folio for a writeback operation:
struct folio *folio = NULL;
while ((folio = writeback_iter(mapping, wbc, folio, &error))) {
error = <do per-foli writeback>;
}The twist here is that the error value is passed by reference, so that the iterator can restore it when breaking out of the loop.
Handling of the magic AOP_WRITEPAGE_ACTIVATE value stays outside the iterator and needs is just kept in the write_cache_pages legacy wrapper. in preparation for eventually killing it off.
Heavily based on a for_each* based iterator from Matthew Wilcox.
--- include/linux/writeback.h | 4 + mm/page-writeback.c | 192 ++++++++++++++++++++++---------------- 2 files changed, 118 insertions(+), 78 deletions(-)
include/linux/writeback.h +4 -0
@@ -82,6 +82,7 @@ struct writeback_control {/* internal fields used by the ->writepages implementation: */ struct folio_batch fbatch;pgoff_t index; +int saved_err;#ifdef CONFIG_CGROUP_WRITEBACK struct bdi_writeback *wb;/* wb this writeback is issued under */
@@ -366,6 +367,9 @@ int balance_dirty_pages_ratelimited_flags(struct address_space *mapping,bool wb_over_bg_thresh (struct bdi_writeback *wb); +struct folio *writeback_iter (struct address_space *mapping, +struct writeback_control *wbc,struct folio *folio,int *error); +typedef int (*writepage_t )(struct folio *folio,struct writeback_control *wbc,void *data);
mm/page-writeback.c +114 -78
234 lines not on this page · raw message
Signature · 1 line
2.39.2
Re: [PATCH 12/13] writeback: add a writeback iterator
Quoted 12 lines · Christoph Hellwig
Refactor the code left in
write_cache_pagesinto an iterator that the file system can call to get the next folio for a writeback operation:struct folio *folio = NULL; while ((folio = writeback_iter(mapping, wbc, folio, &error))) { error = <do per-foli writeback>; }The twist here is that the error value is passed by reference, so that the iterator can restore it when breaking out of the loop.
Quoted 7 lines
Handling of the magic
AOP_WRITEPAGE_ACTIVATEvalue stays outside the iterator and needs is just kept in thewrite_cache_pageslegacy wrapper. in preparation for eventually killing it off.Heavily based on a for_each* based iterator from Matthew Wilcox.
Signed-off-by Christoph Hellwig <[email protected]>
Looks good. Feel free to add:
Honza
Quoted 271 lines
--- include/linux/writeback.h | 4 + mm/page-writeback.c | 192 ++++++++++++++++++++++---------------- 2 files changed, 118 insertions(+), 78 deletions(-)include/linux/writeback.h +4
-0@@ -82,6 +82,7 @@ struct writeback_control {/* internal fields used by the ->writepages implementation: */ struct folio_batch fbatch;pgoff_t index; +int saved_err;#ifdef CONFIG_CGROUP_WRITEBACK struct bdi_writeback *wb;/* wb this writeback is issued under */ @@ -366,6 +367,9 @@ int balance_dirty_pages_ratelimited_flags(struct address_space *mapping,bool wb_over_bg_thresh (struct bdi_writeback *wb); +struct folio *writeback_iter (struct address_space *mapping, +struct writeback_control *wbc,struct folio *folio,int *error); +typedef int (*writepage_t )(struct folio *folio,struct writeback_control *wbc,void *data);mm/page-writeback.c +114
-78@@ -2325,18 +2325,18 @@ void __init page_writeback_init(void) }/** -* tag_pages_for_writeback - tag pages to be written by write_cache_pages +* tag_pages_for_writeback - tag pages to be written by writeback * @mapping: address space structure to write * @start: starting page index * @end: ending page index (inclusive) * * This function scans the page range from @start to @end (inclusive) and tags -* all pages that have DIRTY tag set with a special TOWRITE tag. The idea is -* that write_cache_pages (or whoever calls this function) will then use -* TOWRITE tag to identify pages eligible for writeback. This mechanism is -* used to avoid livelocking of writeback by a process steadily creating new -* dirty pages in the file (thus it is important for this function to be quick -* so that it can tag pages faster than a dirtying process can create them). +* all pages that have DIRTY tag set with a special TOWRITE tag. The caller +* can then use the TOWRITE tag to identify pages eligible for writeback. +* This mechanism is used to avoid livelocking of writeback by a process +* steadily creating new dirty pages in the file (thus it is important for this +* function to be quick so that it can tag pages faster than a dirtying process +* can create them). */ void tag_pages_for_writeback (struct address_space *mapping,pgoff_t start,pgoff_t end)@@ -2434,69 +2434,68 @@ static struct folio *writeback_get_folio(struct address_space *mapping, }/** -* write_cache_pages - walk the list of dirty pages of the given address space and write all of them. +* writeback_iter - iterate folio of a mapping for writeback * @mapping: address space structure to write -* @wbc: subtract the number of written pages from *@wbc->nr_to_write -* @writepage: function called for each page -* @data: data passed to writepage function +* @wbc: writeback context +* @folio: previously iterated folio (%NULL to start) +* @error: in-out pointer for writeback errors (see below) * -* If a page is already under I/O, write_cache_pages() skips it, even -* if it's dirty. This is desirable behaviour for memory-cleaning writeback, -* but it is INCORRECT for data-integrity system calls such as fsync(). fsync() -* and msync() need to guarantee that all the data which was dirty at the time -* the call was made get new I/O started against them. If wbc->sync_mode is -* WB_SYNC_ALL then we were called for data integrity and we must wait for -* existing IO to complete. -* -* To avoid livelocks (when other process dirties new pages), we first tag -* pages which should be written back with TOWRITE tag and only then start -* writing them. For data-integrity sync we have to be careful so that we do -* not miss some pages (e.g., because some other process has cleared TOWRITE -* tag we set). The rule we follow is that TOWRITE tag can be cleared only -* by the process clearing the DIRTY tag (and submitting the page for IO). -* -* To avoid deadlocks between range_cyclic writeback and callers that hold -* pages in PageWriteback to aggregate IO until write_cache_pages() returns, -* we do not loop back to the start of the file. Doing so causes a page -* lock/page writeback access order inversion - we should only ever lock -* multiple pages in ascending page->index order, and looping back to the start -* of the file violates that rule and causes deadlocks. +* This function returns the next folio for the writeback operation described by +* @wbc on @mapping and should be called in a while loop in the ->writepages +* implementation. * -* Return: %0 on success, negative error code otherwise +* To start the writeback operation, %NULL is passed in the @folio argument, and +* for every subsequent iteration the folio returned previously should be passed +* back in. +* +* If there was an error in the per-folio writeback inside the writeback_iter() +* loop, @error should be set to the error value. +* +* Once the writeback described in @wbc has finished, this function will return +* %NULL and if there was an error in any iteration restore it to @error. +* +* Note: callers should not manually break out of the loop using break or goto +* but must keep calling writeback_iter() until it returns %NULL. +* +* Return: the folio to write or %NULL if the loop is done. */ -int write_cache_pages (struct address_space *mapping, -struct writeback_control *wbc,writepage_t writepage, -void *data) +struct folio *writeback_iter (struct address_space *mapping, +struct writeback_control *wbc,struct folio *folio,int *error) { -int ret =0 ; -int error; -struct folio *folio; -pgoff_t end;/* Inclusive */ - -if (wbc->range_cyclic) { - wbc->index = mapping->writeback_index;/* prev offset */ - end =-1 ; - }else { - wbc->index = wbc->range_start >>PAGE_SHIFT ; - end = wbc->range_end >>PAGE_SHIFT ; - } -if (wbc->sync_mode ==WB_SYNC_ALL || wbc->tagged_writepages) -tag_pages_for_writeback (mapping, wbc->index, end); - -folio_batch_init (&wbc->fbatch); +if (!folio) { +folio_batch_init (&wbc->fbatch); + wbc->saved_err = *error =0 ; -for (;;) { - folio =writeback_get_folio (mapping, wbc); -if (!folio) -break ; +/* +* For range cyclic writeback we remember where we stopped so +* that we can continue where we stopped. +* +* For non-cyclic writeback we always start at the beginning of +* the passed in range. +*/ +if (wbc->range_cyclic) + wbc->index = mapping->writeback_index; +else + wbc->index = wbc->range_start >>PAGE_SHIFT ; - error =writepage (folio, wbc, data); +/* +* To avoid livelocks when other processes dirty new pages, we +* first tag pages which should be written back and only then +* start writing them. +* +* For data-integrity writeback we have to be careful so that we +* do not miss some pages (e.g., because some other process has +* cleared the TOWRITE tag we set). The rule we follow is that +* TOWRITE tag can be cleared only by the process clearing the +* DIRTY tag (and submitting the page for I/O). +*/ +if (wbc->sync_mode ==WB_SYNC_ALL || wbc->tagged_writepages) +tag_pages_for_writeback (mapping, wbc->index, +wbc_end (wbc)); + }else { wbc->nr_to_write -=folio_nr_pages (folio); -if (error ==AOP_WRITEPAGE_ACTIVATE ) { -folio_unlock (folio); - error =0 ; - } +WARN_ON_ONCE (*error >0 );/* * For integrity writeback we have to keep going until we have @@ -2510,33 +2509,70 @@ int write_cache_pages(struct address_space *mapping,* wbc->nr_to_write or encounter the first error. */ if (wbc->sync_mode ==WB_SYNC_ALL ) { -if (error && !ret) - ret = error; +if (*error && !wbc->saved_err) + wbc->saved_err = *error; }else { -if (error || wbc->nr_to_write <=0 ) +if (*error || wbc->nr_to_write <=0 ) goto done; } } -/* -* For range cyclic writeback we need to remember where we stopped so -* that we can continue there next time we are called. If we hit the -* last page and there is more work to be done, wrap back to the start -* of the file. -* -* For non-cyclic writeback we always start looking up at the beginning -* of the file if we are called again, which can only happen due to -* -ENOMEM from the file system. -*/ -folio_batch_release (&wbc->fbatch); -if (wbc->range_cyclic) - mapping->writeback_index =0 ; -return ret; + folio =writeback_get_folio (mapping, wbc); +if (!folio) { +/* +* To avoid deadlocks between range_cyclic writeback and callers +* that hold pages in PageWriteback to aggregate I/O until +* the writeback iteration finishes, we do not loop back to the +* start of the file. Doing so causes a page lock/page +* writeback access order inversion - we should only ever lock +* multiple pages in ascending page->index order, and looping +* back to the start of the file violates that rule and causes +* deadlocks. +*/ +if (wbc->range_cyclic) + mapping->writeback_index =0 ; + +/* +* Return the first error we encountered (if there was any) to +* the caller. +*/ + *error = wbc->saved_err; + } +return folio; done:folio_batch_release (&wbc->fbatch);if (wbc->range_cyclic) mapping->writeback_index = folio->index +folio_nr_pages (folio); +return NULL ; +} + +/** +* write_cache_pages - walk the list of dirty pages of the given address space and write all of them. +* @mapping: address space structure to write +* @wbc: subtract the number of written pages from *@wbc->nr_to_write +* @writepage: function called for each page +* @data: data passed to writepage function +* +* Return: %0 on success, negative error code otherwise +* +* Note: please use writeback_iter() instead. +*/ +int write_cache_pages (struct address_space *mapping, +struct writeback_control *wbc,writepage_t writepage, +void *data) +{ +struct folio *folio =NULL ; +int error; + +while ((folio =writeback_iter (mapping, wbc, folio, &error))) { + error =writepage (folio, wbc, data); +if (error ==AOP_WRITEPAGE_ACTIVATE ) { +folio_unlock (folio); + error =0 ; + } + } +return error; }EXPORT_SYMBOL (write_cache_pages);Signature · 1 line
2.39.2
Signature · 2 lines
Jan Kara <[email protected]> SUSE Labs, CR
Re: [PATCH 12/13] writeback: add a writeback iterator
In reply to Christoph HellwigQuoted 24 lines · Christoph Hellwig
Refactor the code left in
write_cache_pagesinto an iterator that the file system can call to get the next folio for a writeback operation:struct folio *folio = NULL; while ((folio = writeback_iter(mapping, wbc, folio, &error))) { error = <do per-foli writeback>; }The twist here is that the error value is passed by reference, so that the iterator can restore it when breaking out of the loop.
Handling of the magic
AOP_WRITEPAGE_ACTIVATEvalue stays outside the iterator and needs is just kept in thewrite_cache_pageslegacy wrapper. in preparation for eventually killing it off.Heavily based on a for_each* based iterator from Matthew Wilcox.
Signed-off-by Christoph Hellwig <[email protected]>--- include/linux/writeback.h | 4 + mm/page-writeback.c | 192 ++++++++++++++++++++++---------------- 2 files changed, 118 insertions(+), 78 deletions(-)
...
Quoted 4 lines
mm/page-writeback.c +0
-0
...
Quoted 4 lines
mm/page-writeback.c +0
-0@@ -2434,69 +2434,68 @@ static struct folio *writeback_get_folio(struct address_space *mapping, }/**
...
Quoted 7 lines
*/
mm/page-writeback.c +2
-3-int write_cache_pages (struct address_space *mapping, -struct writeback_control *wbc,writepage_t writepage, -void *data) +struct folio *writeback_iter (struct address_space *mapping, +struct writeback_control *wbc,struct folio *folio,int *error) {
...
mm/page-writeback.c static struct folio *writeback_get_folio(struct address_space *mapping, · line 2498
Quoted 8 lines
mm/page-writeback.c +2
-4+ }else { wbc->nr_to_write -=folio_nr_pages (folio); -if (error ==AOP_WRITEPAGE_ACTIVATE ) { -folio_unlock (folio); - error =0 ; - } +WARN_ON_ONCE (*error >0 );
Why the warning on writeback error here? It looks like new behavior, but maybe I missed something. Otherwise the factoring LGTM.
Brian
Quoted 95 lines
mm/page-writeback.c +54
-17/* * For integrity writeback we have to keep going until we have @@ -2510,33 +2509,70 @@ int write_cache_pages(struct address_space *mapping,* wbc->nr_to_write or encounter the first error. */ if (wbc->sync_mode ==WB_SYNC_ALL ) { -if (error && !ret) - ret = error; +if (*error && !wbc->saved_err) + wbc->saved_err = *error; }else { -if (error || wbc->nr_to_write <=0 ) +if (*error || wbc->nr_to_write <=0 ) goto done; } } -/* -* For range cyclic writeback we need to remember where we stopped so -* that we can continue there next time we are called. If we hit the -* last page and there is more work to be done, wrap back to the start -* of the file. -* -* For non-cyclic writeback we always start looking up at the beginning -* of the file if we are called again, which can only happen due to -* -ENOMEM from the file system. -*/ -folio_batch_release (&wbc->fbatch); -if (wbc->range_cyclic) - mapping->writeback_index =0 ; -return ret; + folio =writeback_get_folio (mapping, wbc); +if (!folio) { +/* +* To avoid deadlocks between range_cyclic writeback and callers +* that hold pages in PageWriteback to aggregate I/O until +* the writeback iteration finishes, we do not loop back to the +* start of the file. Doing so causes a page lock/page +* writeback access order inversion - we should only ever lock +* multiple pages in ascending page->index order, and looping +* back to the start of the file violates that rule and causes +* deadlocks. +*/ +if (wbc->range_cyclic) + mapping->writeback_index =0 ; + +/* +* Return the first error we encountered (if there was any) to +* the caller. +*/ + *error = wbc->saved_err; + } +return folio; done:folio_batch_release (&wbc->fbatch);if (wbc->range_cyclic) mapping->writeback_index = folio->index +folio_nr_pages (folio); +return NULL ; +} + +/** +* write_cache_pages - walk the list of dirty pages of the given address space and write all of them. +* @mapping: address space structure to write +* @wbc: subtract the number of written pages from *@wbc->nr_to_write +* @writepage: function called for each page +* @data: data passed to writepage function +* +* Return: %0 on success, negative error code otherwise +* +* Note: please use writeback_iter() instead. +*/ +int write_cache_pages (struct address_space *mapping, +struct writeback_control *wbc,writepage_t writepage, +void *data) +{ +struct folio *folio =NULL ; +int error; + +while ((folio =writeback_iter (mapping, wbc, folio, &error))) { + error =writepage (folio, wbc, data); +if (error ==AOP_WRITEPAGE_ACTIVATE ) { +folio_unlock (folio); + error =0 ; + } + } +return error; }EXPORT_SYMBOL (write_cache_pages);Signature · 1 line
2.39.2
Brian Foster
Reviewed-byRe: [PATCH 12/13] writeback: add a writeback iterator
Quoted 47 lines · Brian Foster
Quoted 24 lines · Christoph Hellwig
Refactor the code left in
write_cache_pagesinto an iterator that the file system can call to get the next folio for a writeback operation:struct folio *folio = NULL; while ((folio = writeback_iter(mapping, wbc, folio, &error))) { error = <do per-foli writeback>; }The twist here is that the error value is passed by reference, so that the iterator can restore it when breaking out of the loop.
Handling of the magic
AOP_WRITEPAGE_ACTIVATEvalue stays outside the iterator and needs is just kept in thewrite_cache_pageslegacy wrapper. in preparation for eventually killing it off.Heavily based on a for_each* based iterator from Matthew Wilcox.
Signed-off-by Christoph Hellwig <[email protected]>--- include/linux/writeback.h | 4 + mm/page-writeback.c | 192 ++++++++++++++++++++++---------------- 2 files changed, 118 insertions(+), 78 deletions(-)...
Quoted 4 lines
mm/page-writeback.c +0
-0...
Quoted 4 lines
mm/page-writeback.c +0
-0@@ -2434,69 +2434,68 @@ static struct folio *writeback_get_folio(struct address_space *mapping, }/** ...
Quoted 7 lines
*/
mm/page-writeback.c +2
-3-int write_cache_pages (struct address_space *mapping, -struct writeback_control *wbc,writepage_t writepage, -void *data) +struct folio *writeback_iter (struct address_space *mapping, +struct writeback_control *wbc,struct folio *folio,int *error) {...
Quoted 3 lines
mm/page-writeback.c +1
-0+ }else { wbc->nr_to_write -=folio_nr_pages (folio);
Quoted 8 lines
mm/page-writeback.c static struct folio *writeback_get_folio(struct address_space *mapping, · line 2498
Quoted 5 lines
mm/page-writeback.c +1
-4-if (error ==AOP_WRITEPAGE_ACTIVATE ) { -folio_unlock (folio); - error =0 ; - } +WARN_ON_ONCE (*error >0 );Why the warning on writeback error here? It looks like new behavior, but maybe I missed something. Otherwise the factoring LGTM.
Err, sorry.. I glossed over the > 0 check and read it as < 0. Disregard, this seems reasonable to me as long as we no longer expect those AOP returns (which I'm not really clear on either, but anyways..):
Quoted 100 lines
Brian
Quoted 95 lines
mm/page-writeback.c +54
-17/* * For integrity writeback we have to keep going until we have @@ -2510,33 +2509,70 @@ int write_cache_pages(struct address_space *mapping,* wbc->nr_to_write or encounter the first error. */ if (wbc->sync_mode ==WB_SYNC_ALL ) { -if (error && !ret) - ret = error; +if (*error && !wbc->saved_err) + wbc->saved_err = *error; }else { -if (error || wbc->nr_to_write <=0 ) +if (*error || wbc->nr_to_write <=0 ) goto done; } } -/* -* For range cyclic writeback we need to remember where we stopped so -* that we can continue there next time we are called. If we hit the -* last page and there is more work to be done, wrap back to the start -* of the file. -* -* For non-cyclic writeback we always start looking up at the beginning -* of the file if we are called again, which can only happen due to -* -ENOMEM from the file system. -*/ -folio_batch_release (&wbc->fbatch); -if (wbc->range_cyclic) - mapping->writeback_index =0 ; -return ret; + folio =writeback_get_folio (mapping, wbc); +if (!folio) { +/* +* To avoid deadlocks between range_cyclic writeback and callers +* that hold pages in PageWriteback to aggregate I/O until +* the writeback iteration finishes, we do not loop back to the +* start of the file. Doing so causes a page lock/page +* writeback access order inversion - we should only ever lock +* multiple pages in ascending page->index order, and looping +* back to the start of the file violates that rule and causes +* deadlocks. +*/ +if (wbc->range_cyclic) + mapping->writeback_index =0 ; + +/* +* Return the first error we encountered (if there was any) to +* the caller. +*/ + *error = wbc->saved_err; + } +return folio; done:folio_batch_release (&wbc->fbatch);if (wbc->range_cyclic) mapping->writeback_index = folio->index +folio_nr_pages (folio); +return NULL ; +} + +/** +* write_cache_pages - walk the list of dirty pages of the given address space and write all of them. +* @mapping: address space structure to write +* @wbc: subtract the number of written pages from *@wbc->nr_to_write +* @writepage: function called for each page +* @data: data passed to writepage function +* +* Return: %0 on success, negative error code otherwise +* +* Note: please use writeback_iter() instead. +*/ +int write_cache_pages (struct address_space *mapping, +struct writeback_control *wbc,writepage_t writepage, +void *data) +{ +struct folio *folio =NULL ; +int error; + +while ((folio =writeback_iter (mapping, wbc, folio, &error))) { + error =writepage (folio, wbc, data); +if (error ==AOP_WRITEPAGE_ACTIVATE ) { +folio_unlock (folio); + error =0 ; + } + } +return error; }EXPORT_SYMBOL (write_cache_pages);Signature · 1 line
2.39.2
Jan Kara
maintainerRe: [PATCH 12/13] writeback: add a writeback iterator
Quoted 57 lines · Brian Foster
Quoted 47 lines · Brian Foster
Quoted 24 lines · Christoph Hellwig
Refactor the code left in
write_cache_pagesinto an iterator that the file system can call to get the next folio for a writeback operation:struct folio *folio = NULL; while ((folio = writeback_iter(mapping, wbc, folio, &error))) { error = <do per-foli writeback>; }The twist here is that the error value is passed by reference, so that the iterator can restore it when breaking out of the loop.
Handling of the magic
AOP_WRITEPAGE_ACTIVATEvalue stays outside the iterator and needs is just kept in thewrite_cache_pageslegacy wrapper. in preparation for eventually killing it off.Heavily based on a for_each* based iterator from Matthew Wilcox.
Signed-off-by Christoph Hellwig <[email protected]>--- include/linux/writeback.h | 4 + mm/page-writeback.c | 192 ++++++++++++++++++++++---------------- 2 files changed, 118 insertions(+), 78 deletions(-)...
Quoted 4 lines
mm/page-writeback.c +0
-0...
Quoted 4 lines
mm/page-writeback.c +0
-0@@ -2434,69 +2434,68 @@ static struct folio *writeback_get_folio(struct address_space *mapping, }/** ...
Quoted 7 lines
*/
mm/page-writeback.c +2
-3-int write_cache_pages (struct address_space *mapping, -struct writeback_control *wbc,writepage_t writepage, -void *data) +struct folio *writeback_iter (struct address_space *mapping, +struct writeback_control *wbc,struct folio *folio,int *error) {...
Quoted 3 lines
mm/page-writeback.c +1
-0+ }else { wbc->nr_to_write -=folio_nr_pages (folio);Quoted 8 lines
mm/page-writeback.c static struct folio *writeback_get_folio(struct address_space *mapping, · line 2498
Quoted 5 lines
mm/page-writeback.c +1
-4-if (error ==AOP_WRITEPAGE_ACTIVATE ) { -folio_unlock (folio); - error =0 ; - } +WARN_ON_ONCE (*error >0 );Why the warning on writeback error here? It looks like new behavior, but maybe I missed something. Otherwise the factoring LGTM.
Quoted 5 lines
Err, sorry.. I glossed over the > 0 check and read it as < 0. Disregard, this seems reasonable to me as long as we no longer expect those AOP returns (which I'm not really clear on either, but anyways..):
Reviewed-by Brian Foster <[email protected]>
So my understanding is that AOP_WRITEPAGE_ACTIVATE should be now handled directly by the caller of ->writepage hook and not by writeback_iter() which is the reason why the warning is here.
Honza
Signature · 2 lines
Jan Kara <[email protected]> SUSE Labs, CR
Re: [PATCH 12/13] writeback: add a writeback iterator
Quoted 62 lines · Jan Kara
Quoted 61 lines · Brian Foster
Quoted 47 lines · Brian Foster
Quoted 24 lines · Christoph Hellwig
Refactor the code left in
write_cache_pagesinto an iterator that the file system can call to get the next folio for a writeback operation:struct folio *folio = NULL; while ((folio = writeback_iter(mapping, wbc, folio, &error))) { error = <do per-foli writeback>; }The twist here is that the error value is passed by reference, so that the iterator can restore it when breaking out of the loop.
Handling of the magic
AOP_WRITEPAGE_ACTIVATEvalue stays outside the iterator and needs is just kept in thewrite_cache_pageslegacy wrapper. in preparation for eventually killing it off.Heavily based on a for_each* based iterator from Matthew Wilcox.
Signed-off-by Christoph Hellwig <[email protected]>--- include/linux/writeback.h | 4 + mm/page-writeback.c | 192 ++++++++++++++++++++++---------------- 2 files changed, 118 insertions(+), 78 deletions(-)...
Quoted 4 lines
mm/page-writeback.c +0
-0...
Quoted 4 lines
mm/page-writeback.c +0
-0@@ -2434,69 +2434,68 @@ static struct folio *writeback_get_folio(struct address_space *mapping, }/** ...
Quoted 7 lines
*/
mm/page-writeback.c +2
-3-int write_cache_pages (struct address_space *mapping, -struct writeback_control *wbc,writepage_t writepage, -void *data) +struct folio *writeback_iter (struct address_space *mapping, +struct writeback_control *wbc,struct folio *folio,int *error) {...
Quoted 3 lines
mm/page-writeback.c +1
-0+ }else { wbc->nr_to_write -=folio_nr_pages (folio);Quoted 8 lines
mm/page-writeback.c static struct folio *writeback_get_folio(struct address_space *mapping, · line 2498
Quoted 5 lines
mm/page-writeback.c +1
-4-if (error ==AOP_WRITEPAGE_ACTIVATE ) { -folio_unlock (folio); - error =0 ; - } +WARN_ON_ONCE (*error >0 );Why the warning on writeback error here? It looks like new behavior, but maybe I missed something. Otherwise the factoring LGTM.
Err, sorry.. I glossed over the > 0 check and read it as < 0. Disregard, this seems reasonable to me as long as we no longer expect those AOP returns (which I'm not really clear on either, but anyways..):
Quoted 6 lines
Quoted 1 line
Reviewed-by Brian Foster <[email protected]>So my understanding is that
AOP_WRITEPAGE_ACTIVATEshould be now handled directly by the caller of ->writepage hook and not bywriteback_iter()which is the reason why the warning is here.
Yeah, I wasn't really familiar with the AOP error codes, saw that multiple existed and just assumed they might be arbitrarily relevant across different aop callbacks (and so then filtered the check/clear for WRITEPAGE_ACTIVATE out of my brain ;). On taking a closer look, it seems like the only other one (AOP_TRUNCATED_PAGE) doesn't have any relevance to ->writepage(), so this all makes more sense to me now. Thanks.
Brian
Quoted 5 lines
Honza
Signature · 2 lines
Jan Kara <[email protected]> SUSE Labs, CR
PATCH 13/13 writeback: Remove a use of write_cache_pages() from do_writepages()
In reply to Christoph HellwigFrom: "Matthew Wilcox (Oracle)" <[email protected]>
Use the new writeback_iter() directly instead of indirecting through a callback.
--- mm/page-writeback.c | 31 +++++++++++++++++++------------ 1 file changed, 19 insertions(+), 12 deletions(-)
mm/page-writeback.c +19 -12
@@ -2577,13 +2577,25 @@ int write_cache_pages(struct address_space *mapping, }EXPORT_SYMBOL (write_cache_pages); -static int writepage_cb (struct folio *folio,struct writeback_control *wbc, -void *data) +static int writeback_use_writepage (struct address_space *mapping, +struct writeback_control *wbc) { -struct address_space *mapping = data; -int ret = mapping->a_ops->writepage (&folio->page, wbc); -mapping_set_error (mapping, ret); -return ret; +struct folio *folio =NULL ; +struct blk_plug plug; +int err; + +blk_start_plug (&plug); +while ((folio =writeback_iter (mapping, wbc, folio, &err))) { + err = mapping->a_ops->writepage (&folio->page, wbc); +mapping_set_error (mapping, err); +if (err ==AOP_WRITEPAGE_ACTIVATE ) { +folio_unlock (folio); + err =0 ; + } + } +blk_finish_plug (&plug); + +return err; }int do_writepages (struct address_space *mapping,struct writeback_control *wbc)
@@ -2599,12 +2611,7 @@ int do_writepages(struct address_space *mapping, struct writeback_control *wbc)if (mapping->a_ops->writepages) { ret = mapping->a_ops->writepages (mapping, wbc); }else if (mapping->a_ops->writepage) { -struct blk_plug plug; - -blk_start_plug (&plug); - ret =write_cache_pages (mapping, wbc, writepage_cb, - mapping); -blk_finish_plug (&plug); + ret =writeback_use_writepage (mapping, wbc); }else {/* deal with chardevs and other special files */ ret =0 ;
Signature · 1 line
2.39.2
Re: [PATCH 13/13] writeback: Remove a use of write_cache_pages() from do_writepages()
Quoted 8 lines · Christoph Hellwig
From: "Matthew Wilcox (Oracle)" <[email protected]>
Use the new
writeback_iter()directly instead of indirecting through a callback.Signed-off-by Matthew Wilcox (Oracle) <[email protected]>[hch: ported to the while based iter style]Signed-off-by Christoph Hellwig <[email protected]>
Looks good. Feel free to add:
I've just noticed one preexisting problem which was made more visible by your reshuffling:
Quoted 5 lines
mm/page-writeback.c +2
-2+static int writeback_use_writepage (struct address_space *mapping, +struct writeback_control *wbc) { -struct address_space *mapping = data; -int ret = mapping->a_ops->writepage (&folio->page, wbc);
mm/page-writeback.c int write_cache_pages(struct address_space *mapping, · line 2590
Quoted 10 lines
mm/page-writeback.c +8
-2-mapping_set_error (mapping, ret); -return ret; +struct folio *folio =NULL ; +struct blk_plug plug; +int err; + +blk_start_plug (&plug); +while ((folio =writeback_iter (mapping, wbc, folio, &err))) { + err = mapping->a_ops->writepage (&folio->page, wbc); +mapping_set_error (mapping, err);
So if ->writepage() returns AOP_WRITEPAGE_ACTIVATE, we shouldn't call mapping_set_error() because that's just going to confuse the error tracking.
mm/page-writeback.c int write_cache_pages(struct address_space *mapping, · line 2599
Quoted 9 lines
mm/page-writeback.c +8
-0+if (err ==AOP_WRITEPAGE_ACTIVATE ) { +folio_unlock (folio); + err =0 ; + } + } +blk_finish_plug (&plug); + +return err; }
Honza
Signature · 2 lines
Jan Kara <[email protected]> SUSE Labs, CR
Re: [PATCH 13/13] writeback: Remove a use of write_cache_pages() from do_writepages()
In reply to Christoph HellwigQuoted 9 lines · Christoph Hellwig
From: "Matthew Wilcox (Oracle)" <[email protected]>
Use the new
writeback_iter()directly instead of indirecting through a callback.Signed-off-by Matthew Wilcox (Oracle) <[email protected]>[hch: ported to the while based iter style]Signed-off-by Christoph Hellwig <[email protected]>---
LGTM, modulo Jan's comments:
Quoted 56 lines
mm/page-writeback.c | 31 +++++++++++++++++++------------ 1 file changed, 19 insertions(+), 12 deletions(-)mm/page-writeback.c +19
-12@@ -2577,13 +2577,25 @@ int write_cache_pages(struct address_space *mapping, }EXPORT_SYMBOL (write_cache_pages); -static int writepage_cb (struct folio *folio,struct writeback_control *wbc, -void *data) +static int writeback_use_writepage (struct address_space *mapping, +struct writeback_control *wbc) { -struct address_space *mapping = data; -int ret = mapping->a_ops->writepage (&folio->page, wbc); -mapping_set_error (mapping, ret); -return ret; +struct folio *folio =NULL ; +struct blk_plug plug; +int err; + +blk_start_plug (&plug); +while ((folio =writeback_iter (mapping, wbc, folio, &err))) { + err = mapping->a_ops->writepage (&folio->page, wbc); +mapping_set_error (mapping, err); +if (err ==AOP_WRITEPAGE_ACTIVATE ) { +folio_unlock (folio); + err =0 ; + } + } +blk_finish_plug (&plug); + +return err; }int do_writepages (struct address_space *mapping,struct writeback_control *wbc)@@ -2599,12 +2611,7 @@ int do_writepages(struct address_space *mapping, struct writeback_control *wbc)if (mapping->a_ops->writepages) { ret = mapping->a_ops->writepages (mapping, wbc); }else if (mapping->a_ops->writepage) { -struct blk_plug plug; - -blk_start_plug (&plug); - ret =write_cache_pages (mapping, wbc, writepage_cb, - mapping); -blk_finish_plug (&plug); + ret =writeback_use_writepage (mapping, wbc); }else {/* deal with chardevs and other special files */ ret =0 ;Signature · 1 line
2.39.2