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] [thread-next>] [day] [month] [year] [list]
Message-ID: <5ea287f0-24cb-4ad4-8448-6e397fbf1ec8@redhat.com>
Date: Tue, 6 May 2025 16:24:48 +0200
From: David Hildenbrand <david@...hat.com>
To: Ryan Roberts <ryan.roberts@....com>,
 Andrew Morton <akpm@...ux-foundation.org>,
 "Matthew Wilcox (Oracle)" <willy@...radead.org>,
 Alexander Viro <viro@...iv.linux.org.uk>,
 Christian Brauner <brauner@...nel.org>, Jan Kara <jack@...e.cz>,
 Dave Chinner <david@...morbit.com>, Catalin Marinas
 <catalin.marinas@....com>, Will Deacon <will@...nel.org>,
 Kalesh Singh <kaleshsingh@...gle.com>, Zi Yan <ziy@...dia.com>
Cc: linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
 linux-fsdevel@...r.kernel.org, linux-mm@...ck.org
Subject: Re: [RFC PATCH v4 4/5] mm/readahead: Store folio order in struct
 file_ra_state

On 06.05.25 12:03, Ryan Roberts wrote:
> On 05/05/2025 11:08, David Hildenbrand wrote:
>> On 30.04.25 16:59, Ryan Roberts wrote:
>>> Previously the folio order of the previous readahead request was
>>> inferred from the folio who's readahead marker was hit. But due to the
>>> way we have to round to non-natural boundaries sometimes, this first
>>> folio in the readahead block is often smaller than the preferred order
>>> for that request. This means that for cases where the initial sync
>>> readahead is poorly aligned, the folio order will ramp up much more
>>> slowly.
>>>
>>> So instead, let's store the order in struct file_ra_state so we are not
>>> affected by any required alignment. We previously made enough room in
>>> the struct for a 16 order field. This should be plenty big enough since
>>> we are limited to MAX_PAGECACHE_ORDER anyway, which is certainly never
>>> larger than ~20.
>>>
>>> Since we now pass order in struct file_ra_state, page_cache_ra_order()
>>> no longer needs it's new_order parameter, so let's remove that.
>>>
>>> Worked example:
>>>
>>> Here we are touching pages 17-256 sequentially just as we did in the
>>> previous commit, but now that we are remembering the preferred order
>>> explicitly, we no longer have the slow ramp up problem. Note
>>> specifically that we no longer have 2 rounds (2x ~128K) of order-2
>>> folios:
>>>
>>> TYPE    STARTOFFS     ENDOFFS        SIZE  STARTPG    ENDPG   NRPG  ORDER  RA
>>> -----  ----------  ----------  ----------  -------  -------  -----  -----  --
>>> HOLE   0x00000000  0x00001000        4096        0        1      1
>>> FOLIO  0x00001000  0x00002000        4096        1        2      1      0
>>> FOLIO  0x00002000  0x00003000        4096        2        3      1      0
>>> FOLIO  0x00003000  0x00004000        4096        3        4      1      0
>>> FOLIO  0x00004000  0x00005000        4096        4        5      1      0
>>> FOLIO  0x00005000  0x00006000        4096        5        6      1      0
>>> FOLIO  0x00006000  0x00007000        4096        6        7      1      0
>>> FOLIO  0x00007000  0x00008000        4096        7        8      1      0
>>> FOLIO  0x00008000  0x00009000        4096        8        9      1      0
>>> FOLIO  0x00009000  0x0000a000        4096        9       10      1      0
>>> FOLIO  0x0000a000  0x0000b000        4096       10       11      1      0
>>> FOLIO  0x0000b000  0x0000c000        4096       11       12      1      0
>>> FOLIO  0x0000c000  0x0000d000        4096       12       13      1      0
>>> FOLIO  0x0000d000  0x0000e000        4096       13       14      1      0
>>> FOLIO  0x0000e000  0x0000f000        4096       14       15      1      0
>>> FOLIO  0x0000f000  0x00010000        4096       15       16      1      0
>>> FOLIO  0x00010000  0x00011000        4096       16       17      1      0
>>> FOLIO  0x00011000  0x00012000        4096       17       18      1      0
>>> FOLIO  0x00012000  0x00013000        4096       18       19      1      0
>>> FOLIO  0x00013000  0x00014000        4096       19       20      1      0
>>> FOLIO  0x00014000  0x00015000        4096       20       21      1      0
>>> FOLIO  0x00015000  0x00016000        4096       21       22      1      0
>>> FOLIO  0x00016000  0x00017000        4096       22       23      1      0
>>> FOLIO  0x00017000  0x00018000        4096       23       24      1      0
>>> FOLIO  0x00018000  0x00019000        4096       24       25      1      0
>>> FOLIO  0x00019000  0x0001a000        4096       25       26      1      0
>>> FOLIO  0x0001a000  0x0001b000        4096       26       27      1      0
>>> FOLIO  0x0001b000  0x0001c000        4096       27       28      1      0
>>> FOLIO  0x0001c000  0x0001d000        4096       28       29      1      0
>>> FOLIO  0x0001d000  0x0001e000        4096       29       30      1      0
>>> FOLIO  0x0001e000  0x0001f000        4096       30       31      1      0
>>> FOLIO  0x0001f000  0x00020000        4096       31       32      1      0
>>> FOLIO  0x00020000  0x00021000        4096       32       33      1      0
>>> FOLIO  0x00021000  0x00022000        4096       33       34      1      0
>>> FOLIO  0x00022000  0x00024000        8192       34       36      2      1
>>> FOLIO  0x00024000  0x00028000       16384       36       40      4      2
>>> FOLIO  0x00028000  0x0002c000       16384       40       44      4      2
>>> FOLIO  0x0002c000  0x00030000       16384       44       48      4      2
>>> FOLIO  0x00030000  0x00034000       16384       48       52      4      2
>>> FOLIO  0x00034000  0x00038000       16384       52       56      4      2
>>> FOLIO  0x00038000  0x0003c000       16384       56       60      4      2
>>> FOLIO  0x0003c000  0x00040000       16384       60       64      4      2
>>> FOLIO  0x00040000  0x00050000       65536       64       80     16      4
>>> FOLIO  0x00050000  0x00060000       65536       80       96     16      4
>>> FOLIO  0x00060000  0x00080000      131072       96      128     32      5
>>> FOLIO  0x00080000  0x000a0000      131072      128      160     32      5
>>> FOLIO  0x000a0000  0x000c0000      131072      160      192     32      5
>>> FOLIO  0x000c0000  0x000e0000      131072      192      224     32      5
>>> FOLIO  0x000e0000  0x00100000      131072      224      256     32      5
>>> FOLIO  0x00100000  0x00120000      131072      256      288     32      5
>>> FOLIO  0x00120000  0x00140000      131072      288      320     32      5  Y
>>> HOLE   0x00140000  0x00800000     7077888      320     2048   1728
>>>
>>> Signed-off-by: Ryan Roberts <ryan.roberts@....com>
>>> ---
>>>    include/linux/fs.h |  2 ++
>>>    mm/filemap.c       |  6 ++++--
>>>    mm/internal.h      |  3 +--
>>>    mm/readahead.c     | 18 +++++++++++-------
>>>    4 files changed, 18 insertions(+), 11 deletions(-)
>>>
>>> diff --git a/include/linux/fs.h b/include/linux/fs.h
>>> index 44362bef0010..cde482a7270a 100644
>>> --- a/include/linux/fs.h
>>> +++ b/include/linux/fs.h
>>> @@ -1031,6 +1031,7 @@ struct fown_struct {
>>>     *      and so were/are genuinely "ahead".  Start next readahead when
>>>     *      the first of these pages is accessed.
>>>     * @ra_pages: Maximum size of a readahead request, copied from the bdi.
>>> + * @order: Preferred folio order used for most recent readahead.
>>
>> Looking at other members, and how it relates to the other members, should we
>> call this something like "ra_prev_order" / "prev_ra_order" to distinguish it
>> from !ra members and indicate the "most recent" semantics similar to "prev_pos"?
> 
> As you know, I'm crap at naming, but...
> 
> start, size, async_size and order make up the parameters for the "most recent"
> readahead request. Where "most recent" includes "current" once passed into
> page_cache_ra_order(). The others don't include "ra" or "prev" in their name so
> wasn't sure it was necessary here.
> 
> ra_pages is a bit different; that's not part of the request, it's a (dynamic)
> ceiling to use when creating requests.
> 
> Personally I'd leave it as is, but no strong opinion.

I'm fine with it staying that way; I was merely trying to make sense of 
it all ...


... maybe a better description of the parameters might make the 
semantics easier to grasp.

""most recent" includes "current" once passed into page_cache_ra_order()"

is *really* hard to digest :)

> 
>>
>> Just a thought while digging through this patch ...
>>
>> ...
>>
>>> --- a/mm/filemap.c
>>> +++ b/mm/filemap.c
>>> @@ -3222,7 +3222,8 @@ static struct file *do_sync_mmap_readahead(struct
>>> vm_fault *vmf)
>>>            if (!(vm_flags & VM_RAND_READ))
>>>                ra->size *= 2;
>>>            ra->async_size = HPAGE_PMD_NR;
>>> -        page_cache_ra_order(&ractl, ra, HPAGE_PMD_ORDER);
>>> +        ra->order = HPAGE_PMD_ORDER;
>>> +        page_cache_ra_order(&ractl, ra);
>>>            return fpin;
>>>        }
>>>    #endif
>>> @@ -3258,8 +3259,9 @@ static struct file *do_sync_mmap_readahead(struct
>>> vm_fault *vmf)
>>>        ra->start = max_t(long, 0, vmf->pgoff - ra->ra_pages / 2);
>>>        ra->size = ra->ra_pages;
>>>        ra->async_size = ra->ra_pages / 4;
>>> +    ra->order = 0;
>>>        ractl._index = ra->start;
>>> -    page_cache_ra_order(&ractl, ra, 0);
>>> +    page_cache_ra_order(&ractl, ra);
>>>        return fpin;
>>>    }
>>
>> Why not let page_cache_ra_order() consume the order and update ra->order (or
>> however it will be called :) ) internally?
> 
> You mean continue to pass new_order as a parameter to page_cache_ra_order()? The
> reason I did it the way I'm doing it is because I thought it would be weird for
> the caller of page_cache_ra_order() to set up all the parameters (start, size,
> async_size) of the request except for order...

Agreed. As above, I think we might do better with the description of 
these parameters in general ...

or even document how page_cache_ra_order() acts on these inputs?

-- 
Cheers,

David / dhildenb


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ