Re: [PATCH 1/2] btrfs: Prevent open-coded arithmetic in kmalloc

["Miquel Sabaté Solà" <mssola@mssola.com>]

[-- Attachment #1: Type: text/plain, Size: 1558 bytes --]

David Sterba @ 2025-09-23 09:00 +02:

> On Tue, Sep 23, 2025 at 08:47:35AM +0200, Miquel Sabaté Solà wrote:
>> David Sterba @ 2025-09-23 08:13 +02:
>>
>> > On Mon, Sep 22, 2025 at 02:47:13PM +0200, Miquel Sabaté Solà wrote:
>> >> Hello,
>> >>
>> >> David Sterba @ 2025-09-22 12:28 +02:
>> >>
>> >> > On Fri, Sep 19, 2025 at 04:58:15PM +0200, Miquel Sabaté Solà wrote:
>> >> >> As pointed out in the documentation, calling 'kmalloc' with open-coded
>> >> >> arithmetic can lead to unfortunate overflows and this particular way of
>> >> >> using it has been deprecated. Instead, it's preferred to use
>> >> >> 'kmalloc_array' in cases where it might apply so an overflow check is
>> >> >> performed.
>> >> >
>> >> > So this is an API cleanup and it makes sense to use the checked
>> >> > multiplication but it should be also said that this is not fixing any
>> >> > overflow because in all cases the multipliers are bounded small numbers
>> >> > derived from number of items in leaves/nodes.
>> >>
>> >> Yes, it's just an API cleanup and I don't think it fixes any current bug
>> >> in the code base. So no need to CC stable or anything like that.
>> >
>> > Still the changelog should say explicitly that it's not a bug fix before
>> > somebody assigns a CVE to it because it mentions overflow.
>>
>> Got it! I will submit a v2 and make this more explicit.
>
> No need to, I've updated the changelog at commit time.

Ah, then ignore the v2 patch set I've just sent, which simply changes
the commit log as you suggested.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 897 bytes --]