linux-kernel - Re: [PATCH v3 09/15] block: Add checks to merging of atomic writes

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: <484a449b-5c7e-4766-97d3-36b01c78687c@oracle.com>
Date: Mon, 12 Feb 2024 11:20:48 +0000
From: John Garry <john.g.garry@...cle.com>
To: Nilay Shroff <nilay@...ux.ibm.com>
Cc: axboe@...nel.dk, brauner@...nel.org, bvanassche@....org,
        dchinner@...hat.com, djwong@...nel.org, hch@....de, jack@...e.cz,
        jbongio@...gle.com, jejb@...ux.ibm.com, kbusch@...nel.org,
        linux-block@...r.kernel.org, linux-fsdevel@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-nvme@...ts.infradead.org,
        linux-scsi@...r.kernel.org, linux-xfs@...r.kernel.org,
        martin.petersen@...cle.com, ming.lei@...hat.com, ojaswin@...ux.ibm.com,
        sagi@...mberg.me, tytso@....edu, viro@...iv.linux.org.uk
Subject: Re: [PATCH v3 09/15] block: Add checks to merging of atomic writes


>> +
> 
>> +	imask = ~mask;
> 
>> +
> 
>> +	/* Top bits are different, so crossed a boundary */
> 
>> +	if ((start & imask) != (end & imask))
> 
>> +		return true;
> 
>> +
> 
>> +	return false;
> 
>> +}
> 
>> +
> 

I'm not sure what is going on with your mail client here.

> 
> 
> Shall we ensure here that we don't cross max limit of atomic write supported by
> 
> device? It seems that if the boundary size is not advertized by the device
> 
> (in fact, I have one NVMe drive which has boundary size zero i.e. nabo/nabspf/
> 
> nawupf are all zero but awupf is non-zero) then we (unconditionally) allow
> 
> merging. However it may be possible that post merging the total size of the
> 
> request may exceed the atomic-write-unit-max-size supported by the device and
> 
> if that happens then most probably we would be able to catch it very late in
> 
> the driver code (if the device is NVMe).
> 
> 
> 
> So is it a good idea to validate here whether we could potentially exceed
> 
> the atomic-write-max-unit-size supported by device before we allow merging?

Note that we have atomic_write_max_bytes and atomic_write_max_unit_size, 
and they are not always the same thing.

> 
> In case we exceed the atomic-write-max-unit-size post merge then don't allow
> 
> merging?

We check this elsewhere. I just expanded the normal check for max 
request size to cover atomic writes.

Normally we check that a merged request would not exceed max_sectors 
value, and this max_sectors value can be got from 
blk_queue_get_max_sectors().

So if you check a function like ll_back_merge_fn(), we have a merging 
size check:

	if (blk_rq_sectors(req) + bio_sectors(bio) >
	    blk_rq_get_max_sectors(req, blk_rq_pos(req))) {
		req_set_nomerge(req->q, req);
		return 0;
	}

And here the blk_rq_get_max_sectors() -> blk_queue_get_max_sectors() 
call now also supports atomic writes (see patch #7):

@@ -167,7 +167,16 @@ static inline unsigned get_max_io_size(struct bio *bio,
  {
..

+	if (bio->bi_opf & REQ_ATOMIC)
+		max_sectors = lim->atomic_write_max_sectors;
+	else
+		max_sectors = lim->max_sectors;

Note that we do not allow merging of atomic and non-atomic writes.

Thanks,
John