| Message ID | 160389418807.300137.8222864749005731859.stgit@warthog.procyon.org.uk (mailing list archive) |
|---|---|
| Headers | show |
| Series | AFS fixes [ver #2] | expand |
Matthew Wilcox <willy@infradead.org> wrote: > > > > +{ > > > > + if (PAGE_SIZE - 1 <= __AFS_PAGE_PRIV_MASK) > > > > + return 1; > > > > + else > > > > + return PAGE_SIZE / (__AFS_PAGE_PRIV_MASK + 1); > > > > > > Could this be DIV_ROUND_UP(PAGE_SIZE, __AFS_PAGE_PRIV_MASK + 1); avoiding > > > a conditional? I appreciate it's calculated at compile time today, but > > > it'll be dynamic with THP. > > > > Hmmm - actually, I want a shift size, not a number of bytes as I divide by it > > twice in afs_page_dirty(). > > __AFS_PAGE_PRIV_MASK is a constant though. If your compiler can't > optimise a divide-by-a-constant-power-of-two into a shift-by-a-constant (*), > it's time to get yourself a new compiler. > > (*) assuming that's faster on the CPU it's targetting. I'm dividing by the resolution, which is calculated from the page size - which in a THP world is variable: static inline unsigned long afs_page_dirty(size_t from, size_t to) { size_t res = afs_page_dirty_resolution(); from /= res; /* Round down */ to = (to - 1) / res; /* Round up */ return ((unsigned long)to << __AFS_PAGE_PRIV_SHIFT) | from; } David
Here's a set of fixes for AFS: (1) Fix copy_file_range() to an afs file now returning EINVAL if the splice_write file op isn't supplied. (2) Fix a deref-before-check in afs_unuse_cell(). (3) Fix a use-after-free in afs_xattr_get_acl(). (4) Fix afs to not try to clear PG_writeback when laundering a page. (5) Fix afs to take a ref on a page that it sets PG_private on and to drop that ref when clearing PG_private. (6) Fix a page leak if write_begin() fails. (7) Fix afs_write_begin() to not alter the dirty region info stored in page->private, but rather do this in afs_write_end() instead when we know what we actually changed. (8) Fix afs_invalidatepage() to alter the dirty region info on a page when partial page invalidation occurs so that we don't inadvertantly include a span of zeros that will get written back if a page gets laundered due to a remote 3rd-party induced invalidation. We mustn't, however, reduce the dirty region if the page has been seen to be mapped (ie. we got called through the page_mkwrite vector) as the page might still be mapped and we might lose data if the file is extended again. (9) Fix the dirty region info to have a lower resolution if the size of the page is too large for this to be encoded (e.g. powerpc32 with 64K pages). Note that this might not be the ideal way to handle this, since it may allow some leakage of undirtied zero bytes to the server's copy in the case of a 3rd-party conflict. To aid (8) and (9), two additional patches are included: (*) Wrap the manipulations of the dirty region info stored in page->private into helper functions. (*) Alter the encoding of the dirty region so that the region bounds can be stored with one fewer bit, making a bit available for the indication of mappedness. The patches can be found here: https://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs.git/log/?h=afs-fixes David --- Dan Carpenter (1): afs: Fix a use after free in afs_xattr_get_acl() David Howells (10): afs: Fix copy_file_range() afs: Fix tracing deref-before-check afs: Fix afs_launder_page to not clear PG_writeback afs: Fix to take ref on page when PG_private is set afs: Fix page leak on afs_write_begin() failure afs: Fix where page->private is set during write afs: Wrap page->private manipulations in inline functions afs: Alter dirty range encoding in page->private afs: Fix afs_invalidatepage to adjust the dirty region afs: Fix dirty-region encoding on ppc32 with 64K pages fs/afs/cell.c | 3 +- fs/afs/dir.c | 3 ++ fs/afs/dir_edit.c | 1 + fs/afs/file.c | 74 +++++++++++++++++++++++---- fs/afs/internal.h | 59 ++++++++++++++++++++++ fs/afs/write.c | 100 ++++++++++++++++++++----------------- fs/afs/xattr.c | 2 +- include/linux/page-flags.h | 1 + include/trace/events/afs.h | 20 ++------ 9 files changed, 189 insertions(+), 74 deletions(-)