Re: squashfs performance regression and readahea

[Xiongwei Song <sxwjean@gmail.com>]

Hi Hsin-Yi and Matthew,

With the patch from the attachment on linux 5.10, ran the command as I
mentioned earlier,
got the results below:
1:40.65 (1m + 40.65s)
1:10.12
1:11.10
1:11.47
1:11.59
1:11.94
1:11.86
1:12.04
1:12.21
1:12.06

The performance has improved obviously, but compared to linux 4.18, the
performance is not so good.

Moreover, I wanted to test on linux 5.18. But I think I should revert
9eec1d897139 ("squashfs: provide backing_dev_info in order to disable
read-ahead"),
right?  Otherwise, the patch doesn't work?

Regards,
Xiongwei

On Mon, May 9, 2022 at 3:11 PM Hsin-Yi Wang <hsinyi@chromium.org> wrote:
>
> On Mon, May 9, 2022 at 3:05 PM Hsin-Yi Wang <hsinyi@chromium.org> wrote:
> >
> > On Mon, May 9, 2022 at 12:45 AM Matthew Wilcox <willy@infradead.org> wrote:
> > >
> > > On Sun, May 08, 2022 at 02:46:40PM +0000, Song, Xiongwei wrote:
> > > > I am facing a performance regression on squashfs.  There are many squashfs
> > > > partitions on our board. I am doing the operations below on 90 squashfs
> > > > partitions:
> > >
> > > Hi Xiongwei.  Thanks for the report.  Hsin-Ye and I have been working on
> > > this problem off-list.  Hsin-Ye, could you send the latest version?
> >
> > Hi Matthew,
> >
> > This is the patch of the latest version. Since it's just being
> > reviewed halfway done, I posted the patch here, or should I send it to
> > the list directly?
> >
> The indent seems broken when I pasted it here. Resend in the attachment.
>
> > Thanks
> >
> > From 03558dcaab93ed3c32498eb70c7f2b1382b218d6 Mon Sep 17 00:00:00 2001
> > From: Hsin-Yi Wang <hsinyi@chromium.org>
> > Date: Sun, 10 Oct 2021 21:22:25 +0800
> > Subject: [PATCH] squashfs: implement readahead
> >
> > Implement readahead callback for squashfs. It will read datablocks
> > which cover pages in readahead request. For a few cases it will
> > not mark page as uptodate, including:
> > - file end is 0.
> > - current batch of pages isn't in the same datablock or not enough in a
> >   datablock.
> > Otherwise pages will be marked as uptodate. The unhandled pages will be
> > updated by readpage later.
> >
> > Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
> > ---
> >  fs/squashfs/file.c | 72 +++++++++++++++++++++++++++++++++++++++++++++-
> >  1 file changed, 71 insertions(+), 1 deletion(-)
> >
> > diff --git a/fs/squashfs/file.c b/fs/squashfs/file.c
> > index 89d492916dea..20ec48cf97c5 100644
> > --- a/fs/squashfs/file.c
> > +++ b/fs/squashfs/file.c
> > @@ -39,6 +39,7 @@
> >  #include "squashfs_fs_sb.h"
> >  #include "squashfs_fs_i.h"
> >  #include "squashfs.h"
> > +#include "page_actor.h"
> >
> >  /*
> >   * Locate cache slot in range [offset, index] for specified inode.  If
> > @@ -494,7 +495,76 @@ static int squashfs_readpage(struct file *file,
> > struct page *page)
> >   return 0;
> >  }
> >
> > +static void squashfs_readahead(struct readahead_control *ractl)
> > +{
> > + struct inode *inode = ractl->mapping->host;
> > + struct squashfs_sb_info *msblk = inode->i_sb->s_fs_info;
> > + size_t mask = (1UL << msblk->block_log) - 1;
> > + size_t shift = msblk->block_log - PAGE_SHIFT;
> > + loff_t req_end = readahead_pos(ractl) + readahead_length(ractl);
> > + loff_t start = readahead_pos(ractl) &~ mask;
> > + size_t len = readahead_length(ractl) + readahead_pos(ractl) - start;
> > + struct squashfs_page_actor *actor;
> > + unsigned int nr_pages = 0;
> > + struct page **pages;
> > + u64 block = 0;
> > + int bsize, res, i, index;
> > + int file_end = i_size_read(inode) >> msblk->block_log;
> > + int max_pages = 1UL << shift;
> > +
> > + readahead_expand(ractl, start, (len | mask) + 1);
> > +
> > + if (readahead_pos(ractl) + readahead_length(ractl) < req_end)
> > + return;
> > +
> > + nr_pages = readahead_count(ractl);
> > + pages = kmalloc_array(nr_pages, sizeof(void *), GFP_KERNEL);
> > + if (!pages)
> > + return;
> > +
> > + actor = squashfs_page_actor_init_special(pages, nr_pages, 0);
> > + if (!actor)
> > + goto out;
> > +
> > + for (;;) {
> > + nr_pages = __readahead_batch(ractl, pages, max_pages);
> > + if (!nr_pages)
> > + break;
> > +
> > + if (readahead_pos(ractl) >= i_size_read(inode) ||
> > +     file_end == 0 || nr_pages < max_pages)
> > + goto skip_pages;
> > +
> > + index = pages[0]->index >> shift;
> > + bsize = read_blocklist(inode, index, &block);
> > + res = squashfs_read_data(inode->i_sb, block, bsize, NULL,
> > + actor);
> > +
> > + if (res >=0 && (pages[nr_pages - 1]->index >> shift) == index)
> > + for (i = 0; i < nr_pages; i++)
> > + SetPageUptodate(pages[i]);
> > +
> > + for (i = 0; i < nr_pages; i++) {
> > + unlock_page(pages[i]);
> > + put_page(pages[i]);
> > + }
> > + }
> > +
> > + kfree(actor);
> > + return;
> > +
> > +skip_pages:
> > + for (i = 0; i < nr_pages; i++) {
> > + unlock_page(pages[i]);
> > + put_page(pages[i]);
> > + }
> > +
> > + kfree(actor);
> > +out:
> > + kfree(pages);
> > +}
> >
> >  const struct address_space_operations squashfs_aops = {
> > - .readpage = squashfs_readpage
> > + .readpage = squashfs_readpage,
> > + .readahead = squashfs_readahead
> >  };
> > --
> > 2.31.0