On Thu, Aug 05, 2021 at 10:31:04AM -0700, Darrick J. Wong wrote: > On Tue, Aug 03, 2021 at 08:31:33PM +0100, Matthew Wilcox (Oracle) wrote: > > kmap_atomic() has the side-effect of disabling pagefaults and > > preemption. kmap_local_page() does not do this and is preferred. > > > > Signed-off-by: Matthew Wilcox (Oracle) <willy@xxxxxxxxxxxxx> > > Pretty straightforward. > Reviewed-by: Darrick J. Wong <djwong@xxxxxxxxxx> > > --D > > > --- > > fs/iomap/buffered-io.c | 10 +++++----- > > 1 file changed, 5 insertions(+), 5 deletions(-) > > > > diff --git a/fs/iomap/buffered-io.c b/fs/iomap/buffered-io.c > > index c1c8cd41ea81..8ee0211bea86 100644 > > --- a/fs/iomap/buffered-io.c > > +++ b/fs/iomap/buffered-io.c > > @@ -223,10 +223,10 @@ static int iomap_read_inline_data(struct inode *inode, struct page *page, > > if (poff > 0) > > iomap_page_create(inode, page); > > > > - addr = kmap_atomic(page) + poff; > > + addr = kmap_local_page(page) + poff; Though now that I think about it: Why does iomap_write_actor still use copy_page_from_iter_atomic? Can that be converted to use regular copy_page_from_iter, which at least sometimes uses kmap_local_page? --D > > memcpy(addr, iomap->inline_data, size); > > memset(addr + size, 0, PAGE_SIZE - poff - size); > > - kunmap_atomic(addr); > > + kunmap_local(addr); > > iomap_set_range_uptodate(page, poff, PAGE_SIZE - poff); > > return PAGE_SIZE - poff; > > } > > @@ -682,9 +682,9 @@ static size_t iomap_write_end_inline(struct inode *inode, struct page *page, > > BUG_ON(!iomap_inline_data_valid(iomap)); > > > > flush_dcache_page(page); > > - addr = kmap_atomic(page); > > - memcpy(iomap_inline_data(iomap, pos), addr + pos, copied); > > - kunmap_atomic(addr); > > + addr = kmap_local_page(page) + pos; > > + memcpy(iomap_inline_data(iomap, pos), addr, copied); > > + kunmap_local(addr); > > > > mark_inode_dirty(inode); > > return copied; > > -- > > 2.30.2 > >
