| Message ID | 20230208145335.307287-2-willy@infradead.org (mailing list archive) |
|---|---|
| State | Superseded, archived |
| Headers | show |
| Series | Prevent ->map_pages from sleeping | expand |
On Wed, Feb 08, 2023 at 02:53:33PM +0000, Matthew Wilcox (Oracle) wrote: > XFS doesn't actually need to be holding the XFS_MMAPLOCK_SHARED > to do this, any more than it needs the XFS_MMAPLOCK_SHARED for a > read() that hits in the page cache. Hmm. From commit cd647d5651c0 ("xfs: use MMAPLOCK around filemap_map_pages()"): The page faultround path ->map_pages is implemented in XFS via filemap_map_pages(). This function checks that pages found in page cache lookups have not raced with truncate based invalidation by checking page->mapping is correct and page->index is within EOF. However, we've known for a long time that this is not sufficient to protect against races with invalidations done by operations that do not change EOF. e.g. hole punching and other fallocate() based direct extent manipulations. The way we protect against these races is we wrap the page fault operations in a XFS_MMAPLOCK_SHARED lock so they serialise against fallocate and truncate before calling into the filemap function that processes the fault. Do the same for XFS's ->map_pages implementation to close this potential data corruption issue. How do we prevent faultaround from racing with fallocate and reflink calls that operate below EOF? --D > Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org> > --- > fs/xfs/xfs_file.c | 17 +---------------- > 1 file changed, 1 insertion(+), 16 deletions(-) > > diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c > index 705250f9f90a..528fc538b6b9 100644 > --- a/fs/xfs/xfs_file.c > +++ b/fs/xfs/xfs_file.c > @@ -1388,25 +1388,10 @@ xfs_filemap_pfn_mkwrite( > return __xfs_filemap_fault(vmf, PE_SIZE_PTE, true); > } > > -static vm_fault_t > -xfs_filemap_map_pages( > - struct vm_fault *vmf, > - pgoff_t start_pgoff, > - pgoff_t end_pgoff) > -{ > - struct inode *inode = file_inode(vmf->vma->vm_file); > - vm_fault_t ret; > - > - xfs_ilock(XFS_I(inode), XFS_MMAPLOCK_SHARED); > - ret = filemap_map_pages(vmf, start_pgoff, end_pgoff); > - xfs_iunlock(XFS_I(inode), XFS_MMAPLOCK_SHARED); > - return ret; > -} > - > static const struct vm_operations_struct xfs_file_vm_ops = { > .fault = xfs_filemap_fault, > .huge_fault = xfs_filemap_huge_fault, > - .map_pages = xfs_filemap_map_pages, > + .map_pages = filemap_map_pages, > .page_mkwrite = xfs_filemap_page_mkwrite, > .pfn_mkwrite = xfs_filemap_pfn_mkwrite, > }; > -- > 2.35.1 >
On Wed, Feb 08, 2023 at 08:39:19AM -0800, Darrick J. Wong wrote: > On Wed, Feb 08, 2023 at 02:53:33PM +0000, Matthew Wilcox (Oracle) wrote: > > XFS doesn't actually need to be holding the XFS_MMAPLOCK_SHARED > > to do this, any more than it needs the XFS_MMAPLOCK_SHARED for a > > read() that hits in the page cache. > > Hmm. From commit cd647d5651c0 ("xfs: use MMAPLOCK around > filemap_map_pages()"): > > The page faultround path ->map_pages is implemented in XFS via > filemap_map_pages(). This function checks that pages found in page > cache lookups have not raced with truncate based invalidation by > checking page->mapping is correct and page->index is within EOF. > > However, we've known for a long time that this is not sufficient to > protect against races with invalidations done by operations that do > not change EOF. e.g. hole punching and other fallocate() based > direct extent manipulations. The way we protect against these > races is we wrap the page fault operations in a XFS_MMAPLOCK_SHARED > lock so they serialise against fallocate and truncate before calling > into the filemap function that processes the fault. > > Do the same for XFS's ->map_pages implementation to close this > potential data corruption issue. > > How do we prevent faultaround from racing with fallocate and reflink > calls that operate below EOF? I don't understand the commit message. It'd be nice to have an example of what's insufficient about the protection. If XFS really needs it, it can trylock the semaphore and return 0 if it fails, falling back to the ->fault path. But I don't think XFS actually needs it. The ->map_pages path trylocks the folio, checks the folio->mapping, checks uptodate, then checks beyond EOF (not relevant to hole punch). Then it takes the page table lock and puts the page(s) into the page tables, unlocks the folio and moves on to the next folio. The hole-punch path, like the truncate path, takes the folio lock, unmaps the folio (which will take the page table lock) and removes it from the page cache. So what's the race?
On Wed, Feb 08, 2023 at 05:12:06PM +0000, Matthew Wilcox wrote: > On Wed, Feb 08, 2023 at 08:39:19AM -0800, Darrick J. Wong wrote: > > On Wed, Feb 08, 2023 at 02:53:33PM +0000, Matthew Wilcox (Oracle) wrote: > > > XFS doesn't actually need to be holding the XFS_MMAPLOCK_SHARED > > > to do this, any more than it needs the XFS_MMAPLOCK_SHARED for a > > > read() that hits in the page cache. > > > > Hmm. From commit cd647d5651c0 ("xfs: use MMAPLOCK around > > filemap_map_pages()"): > > > > The page faultround path ->map_pages is implemented in XFS via > > filemap_map_pages(). This function checks that pages found in page > > cache lookups have not raced with truncate based invalidation by > > checking page->mapping is correct and page->index is within EOF. > > > > However, we've known for a long time that this is not sufficient to > > protect against races with invalidations done by operations that do > > not change EOF. e.g. hole punching and other fallocate() based > > direct extent manipulations. The way we protect against these > > races is we wrap the page fault operations in a XFS_MMAPLOCK_SHARED > > lock so they serialise against fallocate and truncate before calling > > into the filemap function that processes the fault. > > > > Do the same for XFS's ->map_pages implementation to close this > > potential data corruption issue. > > > > How do we prevent faultaround from racing with fallocate and reflink > > calls that operate below EOF? > > I don't understand the commit message. It'd be nice to have an example > of what's insufficient about the protection. When this change was made, "insufficient protection" was a reference to the rather well known fact we'd been bugging MM developers about for well over a decade (i.e. since before ->page_mkwrite existed) that the unlocked page invalidation detection hack used everywhere in the page cache code was broken for page invalidation within EOF. i.e. that cannot be correctly detected by (page->mapping == NULL && page->index > EOF) checks. This was a long standing problem, so after a decade of being ignored, the MMAPLOCK was added to XFS to serialise invalidation against page fault based operations. At the time page faults could instantiate page cache pages whilst invalidation operations like truncate_pagecache_range() were running and hence page faults could be instantiating and mapping pages over the range we are trying to invalidate. We were also finding niche syscalls that caused data corruption due to invalidation races (e.g. see xfs_file_fadvise() to avoid readahead vs hole punch races from fadvise(WILLNEED) and readahead() syscalls), so I did an audit to look for any potential interfaces that could race with invalidation. ->map_pages() being called from within the page fault code and having a broken page->index based check for invalidation looked suspect and potentially broken. Hence I slapped the MMAPLOCK around it to stop it from running while a XFS driven page cache invalidation operation was in progress. We work on the principle that when it comes to data corruption vectors, it is far better to err on the side of safety than it is to play fast and loose. fault-around is a perf optimisation, and taking a rwsem in shared mode is not a major increase in overhead for that path, so there was little risk of regressions in adding serialisation just in case there was an as-yet-unknown data corruption vector from that path. Keep in mind this was written before the mm code handled page cache instantiation serialisation sanely via the mapping->invalidation_lock. The mapping->invalidation_lock solves the same issues in a slightly different way, and it may well be that the different implementation means that we don't need to use it in all the places we place the MMAPLOCK in XFS originally. > If XFS really needs it, > it can trylock the semaphore and return 0 if it fails, falling back to > the ->fault path. But I don't think XFS actually needs it. > > The ->map_pages path trylocks the folio, checks the folio->mapping, > checks uptodate, then checks beyond EOF (not relevant to hole punch). > Then it takes the page table lock and puts the page(s) into the page > tables, unlocks the folio and moves on to the next folio. > > The hole-punch path, like the truncate path, takes the folio lock, > unmaps the folio (which will take the page table lock) and removes > it from the page cache. > > So what's the race? Hole punch is a multi-folio operation, so while we are operating on invalidating one folio, another folio in the range we've already invalidated could be instantiated and mapped, leaving mapped up-to-date pages over a range we *require* the page cache to empty. The original MMAPLOCK could not prevent the instantiation of new page cache pages while an invalidation was running, hence we had to block any operation from page faults that instantiated pages into the page cache or operated on the page cache in any way while an invalidation was being run. The mapping->invalidation_lock solved this specific aspect of the problem, so it's entirely possible that we don't have to care about using MMAPLOCK for filemap_map_pages() any more. But I don't know that for certain, I haven't had any time to investigate it in any detail, and when it comes to data corruption vectors I'm not going to change serialisation mechanisms without a decent amount of investigation. I couldn't ever convince myself there wasn't a problem hence the comment in the commit: "Do the same for XFS's ->map_pages implementation to close this potential data corruption issue." Hence if you can explain to me how filemap_map_pages() cannot race against invalidation without holding the mapping->invalidation_lock without potentially leaving stale data in the page cache over the invalidated range (this isn't an XFS specific issue!), then I don't see a problem with removing the MMAPLOCK from this path. Cheers, Dave.
On Thu, Feb 09, 2023 at 08:53:11AM +1100, Dave Chinner wrote: > > If XFS really needs it, > > it can trylock the semaphore and return 0 if it fails, falling back to > > the ->fault path. But I don't think XFS actually needs it. > > > > The ->map_pages path trylocks the folio, checks the folio->mapping, > > checks uptodate, then checks beyond EOF (not relevant to hole punch). > > Then it takes the page table lock and puts the page(s) into the page > > tables, unlocks the folio and moves on to the next folio. > > > > The hole-punch path, like the truncate path, takes the folio lock, > > unmaps the folio (which will take the page table lock) and removes > > it from the page cache. > > > > So what's the race? > > Hole punch is a multi-folio operation, so while we are operating on > invalidating one folio, another folio in the range we've already > invalidated could be instantiated and mapped, leaving mapped > up-to-date pages over a range we *require* the page cache to empty. Nope. ->map_pages is defined to _not_ instantiate new pages. If there are uptodate pages in the page cache, they can be mapped, but missing pages will be skipped, and left to ->fault to bring in.
On Thu, Feb 09, 2023 at 02:44:20AM +0000, Matthew Wilcox wrote: > On Thu, Feb 09, 2023 at 08:53:11AM +1100, Dave Chinner wrote: > > > If XFS really needs it, > > > it can trylock the semaphore and return 0 if it fails, falling back to > > > the ->fault path. But I don't think XFS actually needs it. > > > > > > The ->map_pages path trylocks the folio, checks the folio->mapping, > > > checks uptodate, then checks beyond EOF (not relevant to hole punch). > > > Then it takes the page table lock and puts the page(s) into the page > > > tables, unlocks the folio and moves on to the next folio. > > > > > > The hole-punch path, like the truncate path, takes the folio lock, > > > unmaps the folio (which will take the page table lock) and removes > > > it from the page cache. > > > > > > So what's the race? > > > > Hole punch is a multi-folio operation, so while we are operating on > > invalidating one folio, another folio in the range we've already > > invalidated could be instantiated and mapped, leaving mapped > > up-to-date pages over a range we *require* the page cache to empty. > > Nope. ->map_pages is defined to _not_ instantiate new pages. > If there are uptodate pages in the page cache, they can be mapped, but > missing pages will be skipped, and left to ->fault to bring in. Sure, but *at the time this change was made* other operations could instantiate pages whilst an invalidate was running, and then ->map_pages could also find them and map them whilst that invalidation was still running. i.e. the race conditions that existed before the mapping->invalidate_lock was introduced (ie. we couldn't intercept read page faults instantiating pages in the page cache at all) didn't require ->map_pages to instantiate the page for it to be able to expose incorrect data to userspace when page faults raced with an ongoing invalidation operation. While this may not be able to happen now if everything is using the mapping->invalidate_lock correctly (because read faults are now intercepted before they can instatiate new page cache pages), it doesn't mean it wasn't possible in the past..... -Dave.
On Fri, Feb 10, 2023 at 08:53:58AM +1100, Dave Chinner wrote: > On Thu, Feb 09, 2023 at 02:44:20AM +0000, Matthew Wilcox wrote: > > On Thu, Feb 09, 2023 at 08:53:11AM +1100, Dave Chinner wrote: > > > > If XFS really needs it, > > > > it can trylock the semaphore and return 0 if it fails, falling back to > > > > the ->fault path. But I don't think XFS actually needs it. > > > > > > > > The ->map_pages path trylocks the folio, checks the folio->mapping, > > > > checks uptodate, then checks beyond EOF (not relevant to hole punch). > > > > Then it takes the page table lock and puts the page(s) into the page > > > > tables, unlocks the folio and moves on to the next folio. > > > > > > > > The hole-punch path, like the truncate path, takes the folio lock, > > > > unmaps the folio (which will take the page table lock) and removes > > > > it from the page cache. > > > > > > > > So what's the race? > > > > > > Hole punch is a multi-folio operation, so while we are operating on > > > invalidating one folio, another folio in the range we've already > > > invalidated could be instantiated and mapped, leaving mapped > > > up-to-date pages over a range we *require* the page cache to empty. > > > > Nope. ->map_pages is defined to _not_ instantiate new pages. > > If there are uptodate pages in the page cache, they can be mapped, but > > missing pages will be skipped, and left to ->fault to bring in. > > Sure, but *at the time this change was made* other operations could > instantiate pages whilst an invalidate was running, and then > ->map_pages could also find them and map them whilst that > invalidation was still running. i.e. the race conditions that > existed before the mapping->invalidate_lock was introduced (ie. we > couldn't intercept read page faults instantiating pages in the page > cache at all) didn't require ->map_pages to instantiate the page for > it to be able to expose incorrect data to userspace when page faults > raced with an ongoing invalidation operation. > > While this may not be able to happen now if everything is using the > mapping->invalidate_lock correctly (because read faults are now > intercepted before they can instatiate new page cache pages), it > doesn't mean it wasn't possible in the past..... Sorry, still not getting it. Here's the scenario I think you're talking about. We have three threads (probably in different tasks or they may end up getting synchronized on the page table locks). Thread 1 is calling FALLOC_FL_PUNCH_HOLE over a nice wide range. Thread 2 has the file mmaped and takes a read page fault. Thread 3 also has the file mmaped and also takes a read page fault. Thread 2 calls filemap_map_pages and finds the pages gone. It proceeds to call xfs_filemap_fault() which calls filemap_fault() without taking any XFS locks. filemap_fault() kicks off some readahead which allocates some pages & puts them in the page cache. It calls into xfs_vm_readahead() which calls iomap_readahead() without taking any XFS locks. iomap_readahead() will then call back into xfs_read_iomap_begin() which takes the XFS_ILOCK_SHARED. Since thread 1 is holding XFS_IOLOCK_EXCL, I presume thread 2 will block at this point until thread 1 is done. At this point, the page is still not uptodate, so thread 3 will not map the page if it finds it in >map_pages. Or have I misunderstood XFS inode locking? Entirely possible, it seems quite complicated. Nevertheless, it seems to me that if there's locking that's missing, there's ample opportunities for XFS to take those missing locks in the (slow) fault path, and not take them in the (fast) map_pages path.
On Thu, Feb 09, 2023 at 10:34:21PM +0000, Matthew Wilcox wrote: > On Fri, Feb 10, 2023 at 08:53:58AM +1100, Dave Chinner wrote: > > On Thu, Feb 09, 2023 at 02:44:20AM +0000, Matthew Wilcox wrote: > > > On Thu, Feb 09, 2023 at 08:53:11AM +1100, Dave Chinner wrote: > > > > > If XFS really needs it, > > > > > it can trylock the semaphore and return 0 if it fails, falling back to > > > > > the ->fault path. But I don't think XFS actually needs it. > > > > > > > > > > The ->map_pages path trylocks the folio, checks the folio->mapping, > > > > > checks uptodate, then checks beyond EOF (not relevant to hole punch). > > > > > Then it takes the page table lock and puts the page(s) into the page > > > > > tables, unlocks the folio and moves on to the next folio. > > > > > > > > > > The hole-punch path, like the truncate path, takes the folio lock, > > > > > unmaps the folio (which will take the page table lock) and removes > > > > > it from the page cache. > > > > > > > > > > So what's the race? > > > > > > > > Hole punch is a multi-folio operation, so while we are operating on > > > > invalidating one folio, another folio in the range we've already > > > > invalidated could be instantiated and mapped, leaving mapped > > > > up-to-date pages over a range we *require* the page cache to empty. > > > > > > Nope. ->map_pages is defined to _not_ instantiate new pages. > > > If there are uptodate pages in the page cache, they can be mapped, but > > > missing pages will be skipped, and left to ->fault to bring in. > > > > Sure, but *at the time this change was made* other operations could > > instantiate pages whilst an invalidate was running, and then > > ->map_pages could also find them and map them whilst that > > invalidation was still running. i.e. the race conditions that > > existed before the mapping->invalidate_lock was introduced (ie. we > > couldn't intercept read page faults instantiating pages in the page > > cache at all) didn't require ->map_pages to instantiate the page for > > it to be able to expose incorrect data to userspace when page faults > > raced with an ongoing invalidation operation. > > > > While this may not be able to happen now if everything is using the > > mapping->invalidate_lock correctly (because read faults are now > > intercepted before they can instatiate new page cache pages), it > > doesn't mean it wasn't possible in the past..... > > Sorry, still not getting it. Here's the scenario I think you're > talking about. We have three threads (probably in different tasks > or they may end up getting synchronized on the page table locks). > > Thread 1 is calling FALLOC_FL_PUNCH_HOLE over a nice wide range. > Thread 2 has the file mmaped and takes a read page fault. > Thread 3 also has the file mmaped and also takes a read page fault. > > Thread 2 calls filemap_map_pages and finds the pages gone. It proceeds > to call xfs_filemap_fault() which calls filemap_fault() without > taking any XFS locks. filemap_fault() kicks off some readahead which > allocates some pages & puts them in the page cache. It calls into > xfs_vm_readahead() which calls iomap_readahead() without taking any XFS > locks. iomap_readahead() will then call back into xfs_read_iomap_begin() > which takes the XFS_ILOCK_SHARED. > > Since thread 1 is holding XFS_IOLOCK_EXCL, I presume thread 2 will > block at this point until thread 1 is done. No, because XFS_IOLOCK is not the same lock as XFS_ILOCK. IOLOCK (inode->i_rwsem) and MMAPLOCK(mapping->invalidate_lock) serialise user access to user data (i.e. page cache and direct IO). ILOCK (xfs_inode->i_ilock) serialises access to internal XFS inode metadata such as the extent list. The lock ordering is IOLOCK -> MMAPLOCK -> folio lock -> ILOCK, as described in fs/xfs/xfs_inode.c In the case we are talking about here, operations such as fallocate operate directly on the extent map protected by the ILOCK, so they first have to serialise user access to the data (i.e. take the IOLOCK, MMAPLOCK, run inode_dio_wait() to drain running IOs, run break_layouts() to recall remote pNFS access delegations and serialise DAX accesses, etc), then flush any remaining dirty cached data (which may require allocation and hence taking the ILOCK) and then (maybe) invalidate the cached data over the range that is about to be operated on. Only once we hold all these locks and have performed all these user data operations whilst holding those locks can we then take the ILOCK and directly manipulate the extent map knowing we have locked out all avenues of user data modification whilst we modify the extent map and change the user data contained in the file. > At this point, the page > is still not uptodate, so thread 3 will not map the page if it finds it > in >map_pages. > > Or have I misunderstood XFS inode locking? Entirely possible, it > seems quite complicated. Yes, it is, but as every other filesystem has encountered the same problems that XFS has been dealing with since day zero they've grown exactly the same locking requirements. Some of these are VFS locks (i_rwsem, invalidate_lock) or infrastructure (inode_dio_wait()), but over the long term linux filesystems and the VFS have been trending towards the original XFS locking model that it inherited from Irix 30 years ago, not the other way around. > Nevertheless, it seems to me that if there's > locking that's missing, there's ample opportunities for XFS to take those > missing locks in the (slow) fault path, and not take them in the (fast) > map_pages path. If you go back to the series that introduced the mapping->invalidate_lock and the XFS conversion to use it for the MMAPLOCK in commit 2433480a7e1d ("xfs: Convert to use invalidate_lock"), that's what pulled filemap_fault() out from under the MMAPLOCK. i.e. we used to run the entire fault path under the MMAPLOCK to try to avoid issues with read faults instantiating new pages whilst we were invalidating pages in the same mapping. i.e. we used to serialise both the read fault and write fault path entirely against invalidation. The conversion to the invalidate_lock drove the locking further into the filemap_fault path, so we didn't need to take it for read faults anymore. We still have to the take it for write faults (i.e. ->page_mkwrite) because the page cache is already populated and we still need to serialise ->page_mkwrite against truncate, hole punch, etc. Maybe we didn't need to lock the ->map_pages() path, but after seeing data corruption issues caused by user directed speculative page cache readahead via fadvise() and readahead() syscalls racing with operations that need exclusive invalidation, I didn't think that was a chance worth taking. So, as I've already said, it's entirely possible that we don't need the MMAPLOCK in this path anymore. All I want is a concrete explanation of how the page fault and VFS paths now serialise against invalidation to prevent these historic invalidation race conditions from ever occurring again in the commit message. Cheers, Dave.
diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c index 705250f9f90a..528fc538b6b9 100644 --- a/fs/xfs/xfs_file.c +++ b/fs/xfs/xfs_file.c @@ -1388,25 +1388,10 @@ xfs_filemap_pfn_mkwrite( return __xfs_filemap_fault(vmf, PE_SIZE_PTE, true); } -static vm_fault_t -xfs_filemap_map_pages( - struct vm_fault *vmf, - pgoff_t start_pgoff, - pgoff_t end_pgoff) -{ - struct inode *inode = file_inode(vmf->vma->vm_file); - vm_fault_t ret; - - xfs_ilock(XFS_I(inode), XFS_MMAPLOCK_SHARED); - ret = filemap_map_pages(vmf, start_pgoff, end_pgoff); - xfs_iunlock(XFS_I(inode), XFS_MMAPLOCK_SHARED); - return ret; -} - static const struct vm_operations_struct xfs_file_vm_ops = { .fault = xfs_filemap_fault, .huge_fault = xfs_filemap_huge_fault, - .map_pages = xfs_filemap_map_pages, + .map_pages = filemap_map_pages, .page_mkwrite = xfs_filemap_page_mkwrite, .pfn_mkwrite = xfs_filemap_pfn_mkwrite, };
XFS doesn't actually need to be holding the XFS_MMAPLOCK_SHARED to do this, any more than it needs the XFS_MMAPLOCK_SHARED for a read() that hits in the page cache. Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org> --- fs/xfs/xfs_file.c | 17 +---------------- 1 file changed, 1 insertion(+), 16 deletions(-)