From: Nick Piggin <nickpiggin@yahoo.com.au>
To: Andrew Morton <akpm@osdl.org>
Cc: Jens Axboe <axboe@suse.de>,
linux-kernel <linux-kernel@vger.kernel.org>,
"Chen, Kenneth W" <kenneth.w.chen@intel.com>
Subject: [patch 3/9] no PF_MEMALLOC tinkering
Date: Tue, 12 Apr 2005 22:49:22 +1000 [thread overview]
Message-ID: <425BC3D2.3020909@yahoo.com.au> (raw)
In-Reply-To: <425BC262.1070500@yahoo.com.au>
[-- Attachment #1: Type: text/plain, Size: 32 bytes --]
3/9
--
SUSE Labs, Novell Inc.
[-- Attachment #2: no-PF_MEMALLOC-tinkering.patch --]
[-- Type: text/plain, Size: 3426 bytes --]
PF_MEMALLOC is really not a tool for tinkering. It is pretty specifically
used to prevent recursion into page reclaim, and to prevent low memory
deadlocks.
The mm/swap_state.c code was the only legitimate tinkerer. Its concern
was addressed by the previous patch.
Signed-off-by: Nick Piggin <nickpiggin@yahoo.com.au>
Index: linux-2.6/mm/swap_state.c
===================================================================
--- linux-2.6.orig/mm/swap_state.c 2005-04-12 22:05:44.000000000 +1000
+++ linux-2.6/mm/swap_state.c 2005-04-12 22:26:12.000000000 +1000
@@ -143,7 +143,6 @@ void __delete_from_swap_cache(struct pag
int add_to_swap(struct page * page)
{
swp_entry_t entry;
- int pf_flags;
int err;
if (!PageLocked(page))
@@ -154,30 +153,11 @@ int add_to_swap(struct page * page)
if (!entry.val)
return 0;
- /* Radix-tree node allocations are performing
- * GFP_ATOMIC allocations under PF_MEMALLOC.
- * They can completely exhaust the page allocator.
- *
- * So PF_MEMALLOC is dropped here. This causes the slab
- * allocations to fail earlier, so radix-tree nodes will
- * then be allocated from the mempool reserves.
- *
- * We're still using __GFP_HIGH for radix-tree node
- * allocations, so some of the emergency pools are available,
- * just not all of them.
- */
-
- pf_flags = current->flags;
- current->flags &= ~PF_MEMALLOC;
-
/*
* Add it to the swap cache and mark it dirty
*/
err = __add_to_swap_cache(page, entry, GFP_ATOMIC|__GFP_NOWARN);
- if (pf_flags & PF_MEMALLOC)
- current->flags |= PF_MEMALLOC;
-
switch (err) {
case 0: /* Success */
SetPageUptodate(page);
Index: linux-2.6/drivers/md/dm-crypt.c
===================================================================
--- linux-2.6.orig/drivers/md/dm-crypt.c 2005-04-12 22:05:44.000000000 +1000
+++ linux-2.6/drivers/md/dm-crypt.c 2005-04-12 22:26:12.000000000 +1000
@@ -331,25 +331,14 @@ crypt_alloc_buffer(struct crypt_config *
struct bio *bio;
unsigned int nr_iovecs = (size + PAGE_SIZE - 1) >> PAGE_SHIFT;
int gfp_mask = GFP_NOIO | __GFP_HIGHMEM;
- unsigned long flags = current->flags;
unsigned int i;
- /*
- * Tell VM to act less aggressively and fail earlier.
- * This is not necessary but increases throughput.
- * FIXME: Is this really intelligent?
- */
- current->flags &= ~PF_MEMALLOC;
-
if (base_bio)
bio = bio_clone(base_bio, GFP_NOIO);
else
bio = bio_alloc(GFP_NOIO, nr_iovecs);
- if (!bio) {
- if (flags & PF_MEMALLOC)
- current->flags |= PF_MEMALLOC;
+ if (!bio)
return NULL;
- }
/* if the last bio was not complete, continue where that one ended */
bio->bi_idx = *bio_vec_idx;
@@ -386,9 +375,6 @@ crypt_alloc_buffer(struct crypt_config *
size -= bv->bv_len;
}
- if (flags & PF_MEMALLOC)
- current->flags |= PF_MEMALLOC;
-
if (!bio->bi_size) {
bio_put(bio);
return NULL;
Index: linux-2.6/fs/mpage.c
===================================================================
--- linux-2.6.orig/fs/mpage.c 2005-04-12 22:05:44.000000000 +1000
+++ linux-2.6/fs/mpage.c 2005-04-12 22:26:12.000000000 +1000
@@ -105,11 +105,6 @@ mpage_alloc(struct block_device *bdev,
bio = bio_alloc(gfp_flags, nr_vecs);
- if (bio == NULL && (current->flags & PF_MEMALLOC)) {
- while (!bio && (nr_vecs /= 2))
- bio = bio_alloc(gfp_flags, nr_vecs);
- }
-
if (bio) {
bio->bi_bdev = bdev;
bio->bi_sector = first_sector;
next prev parent reply other threads:[~2005-04-12 13:50 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-04-12 12:43 [patch 0/9] various (mainly mempool fixes and block layer improvements) Nick Piggin
2005-04-12 12:48 ` [patch 1/9] GFP_ZERO fix Nick Piggin
2005-04-12 19:47 ` Andrew Morton
2005-04-13 1:02 ` Nick Piggin
2005-04-14 10:59 ` Manfred Spraul
2005-04-12 12:48 ` [patch 2/9] mempool gfp flag Nick Piggin
2005-04-12 19:50 ` Andrew Morton
2005-04-13 1:03 ` Nick Piggin
2005-04-12 12:49 ` Nick Piggin [this message]
2005-04-12 19:57 ` [patch 3/9] no PF_MEMALLOC tinkering Andrew Morton
2005-04-13 1:13 ` Nick Piggin
2005-04-12 12:49 ` [patch 4/9] blk: no memory barrier Nick Piggin
2005-04-12 12:50 ` [patch 5/9] blk: branch hints Nick Piggin
2005-04-12 12:50 ` [patch 6/9] blk: unplug later Nick Piggin
2005-04-12 19:58 ` Andrew Morton
2005-04-13 1:32 ` Nick Piggin
2005-04-13 10:20 ` Jens Axboe
2005-04-12 12:51 ` [patch 7/9] blk: efficiency improvements Nick Piggin
2005-04-12 12:52 ` [patch 0/9] blk: reduce locking Nick Piggin
2005-04-12 12:53 ` [patch doh/9] mempool simplify alloc Nick Piggin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=425BC3D2.3020909@yahoo.com.au \
--to=nickpiggin@yahoo.com.au \
--cc=akpm@osdl.org \
--cc=axboe@suse.de \
--cc=kenneth.w.chen@intel.com \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox