* [PATCH][2.6.23-rc8-mm2] Fixes to hugetlbfs_read() support
[not found] ` <b040c32a0710031441v3139bd28lce757b2c63796686@mail.gmail.com>
@ 2007-10-04 21:56 ` Badari Pulavarty
2007-10-05 2:27 ` Ken Chen
0 siblings, 1 reply; 2+ messages in thread
From: Badari Pulavarty @ 2007-10-04 21:56 UTC (permalink / raw)
To: Ken Chen
Cc: David Gibson, Andrew Morton, William Lee Irwin III, Adam Litke,
Nishanth Aravamudan, linux-mm
Hi Ken,
Here is the latest against 2.6.23-rc8-mm2. If you are happy with this,
I will ask Andrew to pick it up. I did test this version :)
Thanks,
Badari
Various fixes to hugetlbfs_read() support
- handle HOLES correcty
- clean up weird looking loop condition reported by Ken Chen
- integrated "not updating file offset after read" fix by Ken Chen
- holding i_mutex to prevent any unwanted interactions with truncate
Signed-off-by: Badari Pulavarty <pbadari@us.ibm.com>
fs/hugetlbfs/inode.c | 36 +++++++++++++++++++++++-------------
1 file changed, 23 insertions(+), 13 deletions(-)
Index: linux-2.6.23-rc8/fs/hugetlbfs/inode.c
===================================================================
--- linux-2.6.23-rc8.orig/fs/hugetlbfs/inode.c 2007-09-27 11:32:23.000000000 -0700
+++ linux-2.6.23-rc8/fs/hugetlbfs/inode.c 2007-10-04 16:42:14.000000000 -0700
@@ -228,11 +228,13 @@ static ssize_t hugetlbfs_read(struct fil
struct address_space *mapping = filp->f_mapping;
struct inode *inode = mapping->host;
unsigned long index = *ppos >> HPAGE_SHIFT;
+ unsigned long offset = *ppos & ~HPAGE_MASK;
unsigned long end_index;
loff_t isize;
- unsigned long offset;
ssize_t retval = 0;
+ mutex_lock(&inode->i_mutex);
+
/* validate length */
if (len == 0)
goto out;
@@ -241,7 +243,6 @@ static ssize_t hugetlbfs_read(struct fil
if (!isize)
goto out;
- offset = *ppos & ~HPAGE_MASK;
end_index = (isize - 1) >> HPAGE_SHIFT;
for (;;) {
struct page *page;
@@ -263,18 +264,23 @@ static ssize_t hugetlbfs_read(struct fil
page = find_get_page(mapping, index);
if (unlikely(page == NULL)) {
/*
- * We can't find the page in the cache - bail out ?
+ * We have a HOLE, zero out the user-buffer for the
+ * length of the hole or request.
*/
- goto out;
+ ret = len < nr ? len : nr;
+ if (clear_user(buf, ret))
+ ret = -EFAULT;
+ } else {
+ /*
+ * We have the page, copy it to user space buffer.
+ */
+ ret = hugetlbfs_read_actor(page, offset, buf, len, nr);
}
- /*
- * Ok, we have the page, copy it to user space buffer.
- */
- ret = hugetlbfs_read_actor(page, offset, buf, len, nr);
if (ret < 0) {
if (retval == 0)
retval = ret;
- page_cache_release(page);
+ if (page)
+ page_cache_release(page);
goto out;
}
@@ -284,12 +290,16 @@ static ssize_t hugetlbfs_read(struct fil
index += offset >> HPAGE_SHIFT;
offset &= ~HPAGE_MASK;
- page_cache_release(page);
- if (ret == nr && len)
- continue;
- goto out;
+ if (page)
+ page_cache_release(page);
+
+ /* short read or no more work */
+ if ((ret != nr) || (len == 0))
+ break;
}
out:
+ *ppos = ((loff_t) index << HPAGE_SHIFT) + offset;
+ mutex_unlock(&inode->i_mutex);
return retval;
}
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 2+ messages in thread