From: Jens Axboe <axboe@kernel.dk>
To: Andrew Morton <akpm@linux-foundation.org>,
Xianting Tian <tian.xianting@h3c.com>
Cc: linux-mm@kvack.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] mm: bio_alloc never fails when set GFP_NOIO, GFP_KERNEL
Date: Sun, 25 Oct 2020 13:24:07 -0600 [thread overview]
Message-ID: <53ffff64-da56-7911-ff95-9201476de9e4@kernel.dk> (raw)
In-Reply-To: <20201025120907.699d178b9d27be114e338680@linux-foundation.org>
On 10/25/20 1:09 PM, Andrew Morton wrote:
> On Wed, 21 Oct 2020 11:11:28 +0800 Xianting Tian <tian.xianting@h3c.com> wrote:
>
>> bio_alloc with __GFP_DIRECT_RECLAIM(which is included in GFP_NOIO,
>> GFP_KERNEL) never fails, as stated in the comments of bio_alloc_bioset.
>>
>> So we can remove multiple unneeded null checks of bio_alloc and simplify
>> the code.
>>
>> We have done it in fs/ext4/readpage.c, fs/ext4/page-io.c, fs/direct-io.c,
>> and so forth.
>>
>
> (cc Jens)
>
>> --- a/mm/page_io.c
>> +++ b/mm/page_io.c
>> @@ -30,18 +30,20 @@ static struct bio *get_swap_bio(gfp_t gfp_flags,
>> struct page *page, bio_end_io_t end_io)
>> {
>> struct bio *bio;
>> + struct block_device *bdev;
>>
>> + /*
>> + * bio_alloc will _always_ be able to allocate a bio if
>> + * __GFP_DIRECT_RECLAIM is set, see comments for bio_alloc_bioset().
>> + */
>> bio = bio_alloc(gfp_flags, 1);
>> - if (bio) {
>> - struct block_device *bdev;
>> + bio->bi_iter.bi_sector = map_swap_page(page, &bdev);
>> + bio_set_dev(bio, bdev);
>> + bio->bi_iter.bi_sector <<= PAGE_SHIFT - 9;
>> + bio->bi_end_io = end_io;
>>
>> - bio->bi_iter.bi_sector = map_swap_page(page, &bdev);
>> - bio_set_dev(bio, bdev);
>> - bio->bi_iter.bi_sector <<= PAGE_SHIFT - 9;
>> - bio->bi_end_io = end_io;
>> + bio_add_page(bio, page, thp_size(page), 0);
>>
>> - bio_add_page(bio, page, thp_size(page), 0);
>> - }
>> return bio;
>> }
>>
>> @@ -351,19 +353,13 @@ int __swap_writepage(struct page *page, struct writeback_control *wbc,
>>
>> ret = 0;
>> bio = get_swap_bio(GFP_NOIO, page, end_write_func);
>> - if (bio == NULL) {
>> - set_page_dirty(page);
>> - unlock_page(page);
>> - ret = -ENOMEM;
>> - goto out;
>> - }
>> bio->bi_opf = REQ_OP_WRITE | REQ_SWAP | wbc_to_write_flags(wbc);
>> bio_associate_blkg_from_page(bio, page);
>> count_swpout_vm_event(page);
>> set_page_writeback(page);
>> unlock_page(page);
>> submit_bio(bio);
>> -out:
>> +
>> return ret;
>> }
>>
>> @@ -416,11 +412,6 @@ int swap_readpage(struct page *page, bool synchronous)
>>
>> ret = 0;
>> bio = get_swap_bio(GFP_KERNEL, page, end_swap_bio_read);
>> - if (bio == NULL) {
>> - unlock_page(page);
>> - ret = -ENOMEM;
>> - goto out;
>> - }
>> disk = bio->bi_disk;
>> /*
>> * Keep this task valid during swap readpage because the oom killer may
>
> I'm reluctant to remove these checks - yours is a fairly subtle
> discovery and things might change in the future. Intentionally or
> otherwise!
On the block/bio side, it all boils down to the mempool backing of the
bio allocations. If __GFP_DIRECT_RECLAIM is set for that, then we'll
always wait on allocations for a bio. That is intentional, by design, to
guarantee forward progress. It used to be __GFP_WAIT based, but I guess
that changed at some point...
--
Jens Axboe
prev parent reply other threads:[~2020-10-25 19:24 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-21 3:11 [PATCH] mm: bio_alloc never fails when set GFP_NOIO, GFP_KERNEL Xianting Tian
2020-10-23 16:40 ` Tianxianting
2020-10-25 19:09 ` Andrew Morton
2020-10-25 19:24 ` Jens Axboe [this message]
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=53ffff64-da56-7911-ff95-9201476de9e4@kernel.dk \
--to=axboe@kernel.dk \
--cc=akpm@linux-foundation.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=tian.xianting@h3c.com \
/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;
as well as URLs for NNTP newsgroup(s).