Re: [PATCH] x86/mm, efi: Check for valid image type

2015-07-30 Thread Matt Fleming
On Wed, 29 Jul, at 05:37:52PM, Dave Young wrote: Not sure if BGRT is useful in kexec kernel, it seems not worth to copy it between kernels. IMO just return in case if (efi_setup) is true make sense. Yes, I think it makes sense to skip touching the BGRT on kexec boot. As a side note, we

Re: [PATCH] x86/mm, efi: Check for valid image type

2015-07-30 Thread Josh Triplett
On Thu, Jul 30, 2015 at 06:33:41PM +0200, Sebastian Andrzej Siewior wrote: On 07/29/2015 06:41 PM, Josh Triplett wrote: This is correct. However I miss the point of saving the image in the first place. From what I see is that I have now 272 KiB in memory which are never used again. Is

Re: [PATCH] x86/mm, efi: Check for valid image type

2015-07-30 Thread Sebastian Andrzej Siewior
On 07/29/2015 06:41 PM, Josh Triplett wrote: This is correct. However I miss the point of saving the image in the first place. From what I see is that I have now 272 KiB in memory which are never used again. Is there a usecase why we have it? From the code it looks like we save it during boot

Re: [PATCH] x86/mm, efi: Check for valid image type

2015-07-29 Thread Josh Triplett
On Wed, Jul 29, 2015 at 10:30:51AM +0200, Sebastian Andrzej Siewior wrote: On 07/29/2015 02:10 AM, j...@joshtriplett.org wrote: On Wed, 22 Jul, at 05:32:44PM, Sebastian Andrzej Siewior wrote: now and then. The data behind that pointer changes on each boot because nobody preserves the

Re: [PATCH] x86/mm, efi: Check for valid image type

2015-07-29 Thread Sebastian Andrzej Siewior
On 07/29/2015 02:10 AM, j...@joshtriplett.org wrote: On Wed, 22 Jul, at 05:32:44PM, Sebastian Andrzej Siewior wrote: now and then. The data behind that pointer changes on each boot because nobody preserves the content across kexec. Right. The kernel copies this image precisely because it

Re: [PATCH] x86/mm, efi: Check for valid image type

2015-07-29 Thread Dave Young
Hi, As you pointed out above, a wild pointer could cause a WARN from early_ioremap. We need to never follow the pointer in the first place after a kexec, unless we have some way to know that it's actually valid. So you assume that the information from ACPI is always correct then?

Re: [PATCH] x86/mm, efi: Check for valid image type

2015-07-28 Thread josh
On Tue, Jul 28, 2015 at 09:51:57PM +0100, Matt Fleming wrote: (Pulling in Josh) Thanks, Matt. On Wed, 22 Jul, at 05:32:44PM, Sebastian Andrzej Siewior wrote: I usually see |Ignoring BGRT: failed to allocate memory for image (wanted 264301314 bytes) |Ignoring BGRT: failed to allocate

Re: [PATCH] x86/mm, efi: Check for valid image type

2015-07-28 Thread Matt Fleming
(Pulling in Josh) On Wed, 22 Jul, at 05:32:44PM, Sebastian Andrzej Siewior wrote: I usually see |Ignoring BGRT: failed to allocate memory for image (wanted 264301314 bytes) |Ignoring BGRT: failed to allocate memory for image (wanted 3925872891 bytes) sometimes I get |[ cut