Messages in this thread | | | Date | Tue, 17 Jun 2008 13:25:16 -0300 | From | "Glauber Costa" <> | Subject | Re: [PATCH] x86: merge setup_memory_map with e820 |
| |
On Tue, Jun 17, 2008 at 1:18 PM, Yinghai Lu <yhlu.kernel@gmail.com> wrote: > On Tue, Jun 17, 2008 at 9:04 AM, H. Peter Anvin <hpa@zytor.com> wrote: >> Glauber Costa wrote: >>>> >>>> + boot_params.e820_entries = new_nr; >>>> + if (copy_e820_map(boot_params.e820_map, boot_params.e820_entries) >>>> < 0) { >>>> +#ifdef CONFIG_X86_64 >>>> + early_panic("Cannot find a valid memory map"); >>>> +#else >>> >>> Can't this be made dependant on HIGHMEM instead of X86_64? It seems >>> high-memory dependant anyway. Also, are you sure >>> this code that maps into himem would work with HIGHMEM disabled anyway? > > that is behavoir before merge. > >>> >> >> Uhm... x86-64 and HIGHMEM are mutually exclusive. >> >> Either way, it shouldn't be dependent on anything; there is no reason why >> the #else clause can't be applied to both. > > remove the early_panic?
If this can be done, I'm for it. However, it would be better to leave this patch as is, and throw in another one that just address that, for bisectability purposes. > YH >
-- Glauber Costa. "Free as in Freedom" http://glommer.net
"The less confident you are, the more serious you have to act."
| |