linux-kernel - Re: [PATCH] mm/vmalloc: warn only once when vmalloc detect invalid gfp flags

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: <b7e215ee-fce6-4a7c-995d-8ea6625b1ae8@ixit.cz>
Date: Tue, 18 Nov 2025 01:51:23 +0100
From: David Heidelberg <david@...t.cz>
To: "Vishal Moola (Oracle)" <vishal.moola@...il.com>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
 Uladzislau Rezki <urezki@...il.com>, linux-mm@...ck.org,
 linux-kernel@...r.kernel.org, phone-devel@...r.kernel.org
Subject: Re: [PATCH] mm/vmalloc: warn only once when vmalloc detect invalid
 gfp flags

On 18/11/2025 01:34, Vishal Moola (Oracle) wrote:
> On Tue, Nov 18, 2025 at 01:05:57AM +0100, David Heidelberg via B4 Relay wrote:
>> From: David Heidelberg <david@...t.cz>
>>
>> Without WARN_ONCE, the logs get spammed immediately after the boot,
>> on devices as OnePlus 6T (Snapdragon 845).
> 
> Do you have any snippets of the logs? There'll be a particularly helpful
> section that says "Unexpected gfp:"...

I didn't save it, thou I'll run the kernel with all logs again tomorrow, 
what I remember it was "__GFP_ACCOUNT" in unexpected section.

> 
>> Fixes: 7179b2256315 ("mm/vmalloc: warn on invalid vmalloc gfp flags")
> 
> For some context, this patch begins enforcing the gfp flags passed into
> vmalloc(). If the logs are getting spammed, that means the device is
> calling vmalloc() with some unsupported gfp flag, many many times.

Yes, it was full dmesg buffer just after the boot and I think ongoing.
> 
>> Signed-off-by: David Heidelberg <david@...t.cz>
>> ---
>> I'm not 100% sure this is the right solution, but having WARN_ONCE or
>> rate limited warnings here helps a lot on devices as OnePlus 6 (sdm845).
> 
> Can you check if it still happens on today's mm-new?
> I just sent a new version earlier today that also masks off
> __GFP_ACCOUNT, and had an iteration earlier that added
> __GFP_HARDWALL.

I can try tomorrow, any chance it'll bubble into next-20251118?

Thank you
David

-- 
David Heidelberg