[stable 3.18+] block/sd: Fix device-imposed transfer length limits

[stable 3.18+] block/sd: Fix device-imposed transfer length limits

[Ben Hutchings]

[-- Attachment #1: Type: text/plain, Size: 495 bytes --]

Please pick the following commit for 3.18-stable and later stable
branches:

commit ca369d51b3e1649be4a72addd6d6a168cfb3f537
Author: Martin K. Petersen <martin.petersen@oracle.com>
Date:   Fri Nov 13 16:46:48 2015 -0500

    block/sd: Fix device-imposed transfer length limits

It fixes a regression that was introduced in 4.2 and then backported to
those stable branches.

Ben.

-- 
Ben Hutchings
All the simple programs have been written, and all the good names taken.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 811 bytes --]

Re: [stable 3.18+] block/sd: Fix device-imposed transfer length limits

[Holger Hoffstätte]

On Fri, 01 Jan 2016 19:51:35 +0000, Ben Hutchings wrote:

> Please pick the following commit for 3.18-stable and later stable
> branches:
> 
> commit ca369d51b3e1649be4a72addd6d6a168cfb3f537
> Author: Martin K. Petersen <martin.petersen@oracle.com>
> Date:   Fri Nov 13 16:46:48 2015 -0500
> 
>     block/sd: Fix device-imposed transfer length limits
> 
> It fixes a regression that was introduced in 4.2 and then backported to
> those stable branches.

Just a heads-up that this will require some additional backporting love,
since it breaks compilation of older md due to (I think) commit
8ae126660fddbe ("block: kill merge_bvec_fn() completely").

I tried to merge this into my 4.1++ (because SMR drama) but offhand
didn't know how to fix the conflict in md, so maybe someone more familiar
with the block layer/bio merging can help.

-h