| Message ID | 4AD223C3.3020607@canonical.com |
|---|---|
| State | Accepted |
| Headers | show |
Tim Gardner wrote: > Andy, Stefan - Why _is_ it that we don't have PAT enabled? Wasn't it > originally a prerequisite for KMS back in Jaunty days? > > There are some comments on LKML pointing out that we _should_ have PAT > enabled, e.g., "[RFC Patch] use MTRR for write combining if PAT is not > available" > > rtg > http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-karmic.git;a=commit;h=a611fa50121f1698bd367b991f66e2268eb10824
On Tue, Oct 13, 2009 at 2:18 AM, Tim Gardner <tim.gardner@canonical.com> wrote: > Tim Gardner wrote: >> Andy, Stefan - Why _is_ it that we don't have PAT enabled? Wasn't it >> originally a prerequisite for KMS back in Jaunty days? >> >> There are some comments on LKML pointing out that we _should_ have PAT >> enabled, e.g., "[RFC Patch] use MTRR for write combining if PAT is not >> available" >> >> rtg >> > > http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-karmic.git;a=commit;h=a611fa50121f1698bd367b991f66e2268eb10824 Excellent. Is the idea that this will be a karmic beta update in a couple of days, and we have a get-out-of-jail of disabling it (or default boot with 'nopat') before the final release is cut? Or, is this simply too late to make the final release? Daniel
Daniel J Blueman wrote: > On Tue, Oct 13, 2009 at 2:18 AM, Tim Gardner <tim.gardner@canonical.com> wrote: >> Tim Gardner wrote: >>> Andy, Stefan - Why _is_ it that we don't have PAT enabled? Wasn't it >>> originally a prerequisite for KMS back in Jaunty days? >>> >>> There are some comments on LKML pointing out that we _should_ have PAT >>> enabled, e.g., "[RFC Patch] use MTRR for write combining if PAT is not >>> available" >>> >>> rtg >>> >> http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-karmic.git;a=commit;h=a611fa50121f1698bd367b991f66e2268eb10824 > > Excellent. Is the idea that this will be a karmic beta update in a > couple of days, and we have a get-out-of-jail of disabling it (or > default boot with 'nopat') before the final release is cut? > > Or, is this simply too late to make the final release? > > Daniel Kernel freeze is Thursday. With the addition of stable and other miscellaneous patches, today's upload is going to be the last one (unless some OMG kitten killer gets in my face). rtg
On Tue, Oct 13, 2009 at 2:10 PM, Tim Gardner <tim.gardner@canonical.com> wrote: > Daniel J Blueman wrote: >> On Tue, Oct 13, 2009 at 2:18 AM, Tim Gardner <tim.gardner@canonical.com> wrote: >>> Tim Gardner wrote: >>>> Andy, Stefan - Why _is_ it that we don't have PAT enabled? Wasn't it >>>> originally a prerequisite for KMS back in Jaunty days? >>>> >>>> There are some comments on LKML pointing out that we _should_ have PAT >>>> enabled, e.g., "[RFC Patch] use MTRR for write combining if PAT is not >>>> available" >>>> >>>> rtg >>>> >>> http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-karmic.git;a=commit;h=a611fa50121f1698bd367b991f66e2268eb10824 >> >> Excellent. Is the idea that this will be a karmic beta update in a >> couple of days, and we have a get-out-of-jail of disabling it (or >> default boot with 'nopat') before the final release is cut? >> >> Or, is this simply too late to make the final release? >> >> Daniel > > Kernel freeze is Thursday. With the addition of stable and other > miscellaneous patches, today's upload is going to be the last one > (unless some OMG kitten killer gets in my face). Ok, great. Still of importance, so worth mentioning here, CONFIG_X86_MCE gives visibility of otherwise silent memory controller and/or internal processor state (including processor passing critical temperature thresholds). This is particularly crucial for servers with lots of memory (therefore higher chance of cell failure), but also modern desktops having larger amounts of memory. From a support perspective, detected memory errors are logged to the kernel log. From a user perspective (particularly for ubuntu-server), the user knows more when they have eg failing/faulty memory. I'd think twice about deploying critical services on a server where (un)correctable ECC errors *silently* occur... Since this feature has been around really a long time, it carries no risk. Thoughts?
Daniel J Blueman wrote: > On Tue, Oct 13, 2009 at 2:10 PM, Tim Gardner <tim.gardner@canonical.com> wrote: >> Daniel J Blueman wrote: >>> On Tue, Oct 13, 2009 at 2:18 AM, Tim Gardner <tim.gardner@canonical.com> wrote: >>>> Tim Gardner wrote: >>>>> Andy, Stefan - Why _is_ it that we don't have PAT enabled? Wasn't it >>>>> originally a prerequisite for KMS back in Jaunty days? >>>>> >>>>> There are some comments on LKML pointing out that we _should_ have PAT >>>>> enabled, e.g., "[RFC Patch] use MTRR for write combining if PAT is not >>>>> available" >>>>> >>>>> rtg >>>>> >>>> http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-karmic.git;a=commit;h=a611fa50121f1698bd367b991f66e2268eb10824 >>> Excellent. Is the idea that this will be a karmic beta update in a >>> couple of days, and we have a get-out-of-jail of disabling it (or >>> default boot with 'nopat') before the final release is cut? >>> >>> Or, is this simply too late to make the final release? >>> >>> Daniel >> Kernel freeze is Thursday. With the addition of stable and other >> miscellaneous patches, today's upload is going to be the last one >> (unless some OMG kitten killer gets in my face). > > Ok, great. > > Still of importance, so worth mentioning here, CONFIG_X86_MCE gives > visibility of otherwise silent memory controller and/or internal > processor state (including processor passing critical temperature > thresholds). > > This is particularly crucial for servers with lots of memory > (therefore higher chance of cell failure), but also modern desktops > having larger amounts of memory. From a support perspective, detected > memory errors are logged to the kernel log. From a user perspective > (particularly for ubuntu-server), the user knows more when they have > eg failing/faulty memory. I'd think twice about deploying critical > services on a server where (un)correctable ECC errors *silently* > occur... > > Since this feature has been around really a long time, it carries no > risk. Thoughts? Frankly, I've no idea why it _isn't_ enabled. Since Karmic is really the test release for Lucid, I'm gonna turn it on. http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-karmic.git;a=commit;h=ecae83a7adca974982267a4084a30033de9b6b22 I'm not too concerned about MCE working correctly on 64 bit platforms, but I've a bit more concern about older 32 bit CPUs. rtg
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 4427956..2fbc3c6 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -1327,7 +1327,9 @@ config MATH_EMULATION kernel, it won't hurt. config MTRR - bool "MTRR (Memory Type Range Register) support" + bool + default y + prompt "MTRR (Memory Type Range Register) support" if EMBEDDED ---help--- On Intel P6 family processors (Pentium Pro, Pentium II and later) the Memory Type Range Registers (MTRRs) may be used to control @@ -1393,7 +1395,8 @@ config MTRR_SANITIZER_SPARE_REG_NR_DEFAULT config X86_PAT bool - prompt "x86 PAT support" + default y + prompt "x86 PAT support" if EMBEDDED depends on MTRR ---help--- Use PAT attributes to setup page level cache control.