linux-kernel - Re: [RFC PATCH] block: set bdi congested when no tag available

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <CAGWkznEtMQ67G5ebOsrDqVKdXcVZNHA6rLNz0B9jMODkABiSFQ@mail.gmail.com>
Date:   Wed, 15 Jun 2022 16:58:03 +0800
From:   Zhaoyang Huang <huangzhaoyang@...il.com>
To:     Vlastimil Babka <vbabka@...e.cz>
Cc:     Jens Axboe <axboe@...nel.dk>,
        "Martin K. Petersen" <martin.petersen@...cle.com>,
        Ming Lei <ming.lei@...hat.com>, Hannes Reinecke <hare@...e.de>,
        Bart Van Assche <bvanassche@....org>,
        Zhaoyang Huang <zhaoyang.huang@...soc.com>,
        LKML <linux-kernel@...r.kernel.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        "open list:MEMORY MANAGEMENT" <linux-mm@...ck.org>,
        Mel Gorman <mgorman@...hsingularity.net>
Subject: Re: [RFC PATCH] block: set bdi congested when no tag available

On Wed, Mar 9, 2022 at 7:50 PM Vlastimil Babka <vbabka@...e.cz> wrote:
>
> On 3/9/22 08:40, Zhaoyang Huang wrote:
> > From: Zhaoyang Huang <zhaoyang.huang@...soc.com>
> >
> > In previous version, block layer will set bdi to be congested when
> > get_request fail, which may throttle direct_reclaim. Move them back
> > under current blk-mq design.
>
> Hm I thought Mel removed the direct reclaim throttling based on bdi
> congestion in his series [1]. Maybe block layers has other uses (I have no
> idea), for tracking congestion, that should be mentioned instead?
>
> [1]
> https://lore.kernel.org/all/20211022144651.19914-1-mgorman@techsingularity.net/
>
As my understanding, the above patch change suspending on bdi
congested for a given interval to be woken up by finish of request or
bio. This patch is not against above one, which add one more scenario
for setting bdi as congested.
> >
> > Signed-off-by: Zhaoyang Huang <zhaoyang.huang@...soc.com>
> > ---
> >  block/blk-mq-tag.c      | 18 +++++++++++++++++-
> >  include/linux/sbitmap.h |  1 +
> >  lib/sbitmap.c           | 17 +++++++++++++++++
> >  3 files changed, 35 insertions(+), 1 deletion(-)
> >
> > diff --git a/block/blk-mq-tag.c b/block/blk-mq-tag.c
> > index 995336a..bd1e520 100644
> > --- a/block/blk-mq-tag.c
> > +++ b/block/blk-mq-tag.c
> > @@ -11,6 +11,7 @@
> >
> >  #include <linux/blk-mq.h>
> >  #include <linux/delay.h>
> > +#include <linux/backing-dev.h>
> >  #include "blk.h"
> >  #include "blk-mq.h"
> >  #include "blk-mq-sched.h"
> > @@ -126,8 +127,11 @@ unsigned int blk_mq_get_tag(struct blk_mq_alloc_data *data)
> >   if (tag != BLK_MQ_NO_TAG)
> >   goto found_tag;
> >
> > - if (data->flags & BLK_MQ_REQ_NOWAIT)
> > + if (data->flags & BLK_MQ_REQ_NOWAIT) {
> > + set_bdi_congested(data->q->disk->bdi,BLK_RW_SYNC);
> > + set_bdi_congested(data->q->disk->bdi,BLK_RW_ASYNC);
> >   return BLK_MQ_NO_TAG;
> > + }
> >
> >   ws = bt_wait_ptr(bt, data->hctx);
> >   do {
> > @@ -193,9 +197,21 @@ unsigned int blk_mq_get_tag(struct blk_mq_alloc_data *data)
> >   return tag + tag_offset;
> >  }
> >
> > +static bool blk_mq_has_free_tags(struct blk_mq_tags *tags)
> > +{
> > + if (!tags)
> > + return true;
> > +
> > + return sbitmap_any_bit_clear(&tags->bitmap_tags.sb);
> > +}
> > +
> >  void blk_mq_put_tag(struct blk_mq_tags *tags, struct blk_mq_ctx *ctx,
> >       unsigned int tag)
> >  {
> > + if (!blk_mq_has_free_tags(tags)) {
> > + clear_bdi_congested(ctx->queue->disk->bdi, BLK_RW_SYNC);
> > + clear_bdi_congested(ctx->queue->disk->bdi, BLK_RW_ASYNC);
> > + }
> >   if (!blk_mq_tag_is_reserved(tags, tag)) {
> >   const int real_tag = tag - tags->nr_reserved_tags;
> >
> > diff --git a/include/linux/sbitmap.h b/include/linux/sbitmap.h
> > index 4a6ff27..30a5553 100644
> > --- a/include/linux/sbitmap.h
> > +++ b/include/linux/sbitmap.h
> > @@ -210,6 +210,7 @@ static inline void sbitmap_free(struct sbitmap *sb)
> >   * Return: true if any bit in the bitmap is set, false otherwise.
> >   */
> >  bool sbitmap_any_bit_set(const struct sbitmap *sb);
> > +bool sbitmap_any_bit_clear(const struct sbitmap *sb);
> >
> >  #define SB_NR_TO_INDEX(sb, bitnr) ((bitnr) >> (sb)->shift)
> >  #define SB_NR_TO_BIT(sb, bitnr) ((bitnr) & ((1U << (sb)->shift) - 1U))
> > diff --git a/lib/sbitmap.c b/lib/sbitmap.c
> > index 2709ab8..baa30d8 100644
> > --- a/lib/sbitmap.c
> > +++ b/lib/sbitmap.c
> > @@ -309,6 +309,23 @@ bool sbitmap_any_bit_set(const struct sbitmap *sb)
> >  }
> >  EXPORT_SYMBOL_GPL(sbitmap_any_bit_set);
> >
> > +bool sbitmap_any_bit_clear(const struct sbitmap *sb)
> > +{
> > + unsigned int i;
> > +
> > + for (i = 0; i < sb->map_nr; i++) {
> > + const struct sbitmap_word *word = &sb->map[i];
> > + unsigned long mask = word->word & ~word->cleared;
> > + unsigned long ret;
> > +
> > + ret = find_first_zero_bit(&mask, word->depth);
> > + if (ret < word->depth)
> > + return true;
> > + }
> > + return false;
> > +}
> > +EXPORT_SYMBOL_GPL(sbitmap_any_bit_clear);
> > +
> >  static unsigned int __sbitmap_weight(const struct sbitmap *sb, bool set)
> >  {
> >   unsigned int i, weight = 0;
>