Re: kern/54800: pool memory corruption

[Maxime Villard <max%m00nbsd.net@localhost>]

Le 27/12/2019 à 02:20, Andrew Doran a écrit :
> Max,
> 
> On Thu, Dec 26, 2019 at 09:05:00AM +0000, martin%NetBSD.org@localhost wrote:
> 
>> Hidden in the log of the currently failing qemu / anita sparc64 tesst
>> runs you can find this gem:
>>
>> [ 422.4170018] pool_redzone_check: 0x1257a0c20: 0x00 != 0x66
>> [ 422.4170018] pool_redzone_check: 0x1257a0c21: 0x00 != 0x05
>>
>> If I read that correctly, a single 16bit halfword supposed to be 0x0f66
>> has been overwritten with 0.
>>
>> Side question: why is that not worth a panic?
> 
> In rev 1.225 of subr_pool.c you changed this from a panic to a printf.
> It looks like we can make this a panic again.  Do you agree?
> 
> Andrew

(Have to leave right now but will look at and fix it tonight)