... Molnar wrote: > Linus, please pull the x86-pat git tree from: ... /linux-2.6-x86-pat.git for-linus > > this adds the ... final) phase of the x86 PAT changes. Due to > generic impact ... separate tree. So why is PAT dependent on NONPROMISC_DEVMEM here? It ...

linux-kernel - Linus Torvalds - Apr 25 2008 - 19:43

... Molnar wrote: > Linus, please pull the x86-pat git tree from: ... kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6 ... final) phase of the x86 PAT changes. Due to > generic impact ... CONFIG_NONPROMISC_DEVMEM defaults to disabled so PAT will be > disabled, to stay ...

linux-kernel - Gabriel C - Apr 26 2008 - 09:40

... supported > only thanks to VM changes that made it possible. > > But ... neither /dev/mem restrictions nor PAT - paranoia: one less degree of ... take into account Ingo -----------------------> Subject: x86 PAT: decouple from nonpromisc devmem ...

linux-kernel - Ingo Molnar - Apr 26 2008 - 04:56

... should be pr_debug() instead. The idea in pat.c is to only print out ... up the patch below. Ingo --------> Subject: x86 PAT: tone down debugging messages From: ... mm/pat.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) Index: linux-x86. ...

linux-kernel - Ingo Molnar - Apr 26 2008 - 05:57

... isn't true. When disabling PAT again ( no changes to BIOS ) > kvm-intel ... able to disable NONPROMISC_DEVMEM without disabling PAT. Ingo diff --git a/arch/x86/ ... config X86_PAT - def_bool y + bool prompt "x86 PAT support" - depends on MTRR && ...

linux-kernel - Ingo Molnar - Apr 26 2008 - 10:42

... with Ingo's patch that removes PAT dependency on NONPROMISC_DEVMEM. It makes things ... intel.com> --- arch/x86/mm/pat.c | 31 +++++++++++++++++++++++++++++++ 1 file changed, 31 insertions ... #include pat.h> #include ...

linux-kernel - Venki Pallipadi - Apr 26 2008 - 14:32

... ; Thomas Gleixner >Subject: Re: [git pull] x86 PAT changes > > >* Linus Torvalds < ... got supported >> only thanks to VM changes that made it possible. >> >> But read/ ... be no architectural need to make PAT >dependent on nonpromisc-devmem, and thus ...

linux-kernel - Pallipadi, Venkatesh - Apr 26 2008 - 12:54

Linus Torvalds wrote: > > So why is PAT dependent on NONPROMISC_DEVMEM here? > > It seems pointless and wrong. Somebody might want to have both the full > /dev/mem and PAT. > The problem is that that can create cached/uncached aliases, ...

linux-kernel - H. Peter Anvin - Apr 25 2008 - 20:06

... t got time to test 32bit now ) kernel[1] >> with PAT enabled , kvm-intel does not work anymore. >> >> When ... >> BIOS which isn't true. When disabling PAT again ( no changes to BIOS ) >> kvm-intel works again here. >> >> Is that an known problem ? ...

linux-kernel - Gabriel C - Apr 26 2008 - 11:37

... t got time to test 32bit now ) kernel[1] > >> with PAT enabled , kvm-intel does not work anymore. > >> > ... > >> BIOS which isn't true. When disabling PAT again ( no changes to BIOS ) > >> kvm-intel works again here. > >> > >> Is that an known problem ...

linux-kernel - Ingo Molnar - Apr 26 2008 - 11:41

... () on real RAM (because the page counting would get confused), and that actually got supported only thanks to VM changes that made it possible. But read/write has always been supported, and shouldn't cause any cached/uncached aliases! Linus ...

linux-kernel - Linus Torvalds - Apr 25 2008 - 21:12

On Sat, 26 Apr 2008, Ingo Molnar wrote: > * Gabriel C wrote: > > > > It seems like the update triggers some compiler bug. > > > > Setting CC_OPTIMIZE_FOR_SIZE=n fixed that btw. > > > > I don't have the time right

linux-kernel - Linus Torvalds - Apr 26 2008 - 12:43

On Sat, 26 Apr 2008, Pallipadi, Venkatesh wrote: > > Agreed that NONPROMISC_DEVMEM is not really needed for read/write. But, > we will still need it for /dev/mem. If so, just disable it unconditionally for mmap. As mentioned, that's really

linux-kernel - Linus Torvalds - Apr 26 2008 - 13:15