[Karmic] SRU-fix: Fix nx_enable reporting

Tim Gardner tim.gardner at canonical.com
Tue Nov 10 13:46:19 UTC 2009

Amit Kucheria wrote:
> On 09 Nov 10, Colin King wrote:
>> Hi,
>> On Tue, 2009-11-10 at 13:43 +0100, Stefan Bader wrote:
>>> As described in the patch, we got a regression in proposed which was
>>> caused by a SRU update. Though fixing the message fir 64bit it caused
>>> 32bit non-PAE suddenly to believe it had nx protection. Which then
>>> caused sleep problems (for the computer, not the programmer :)).
>>> I guess after being involved in testing, I should get two other ACKs
>>> for it and then respin, proposed. Thanks.
>> This is quite subtle, since _PAGE_NX is non-zero for CONFIG_X86_64 ||
>> CONFIG_X86_PAE configurations, hence zero for 32 bit systems without
>> PAE.  The effect of setting nx_enabled to zero for 32 bit systems
>> without PAE now means that header->pmode_efer_low is now set correctly
>> in acpi_save_state_mem() (sleep.c) for these systems, hence fixing the
>> suspend/resume issue.
>> So, functionally, yep, it looks OK.
>> Acked-by: Colin King <colin.king at canonical.com>
> I understood the subtlety after reading Colin's explanation. I believe his
> explanation should be part of the patch since it better explains the problem.
> Conditionally-Acked-by: Amit Kucheria <amit.kucheria at canonical.com>

Yeah, what Amit said.

Tim Gardner tim.gardner at canonical.com

More information about the kernel-team mailing list