When the TSC is calibrated against the PIT due to the nonavailability
of PMTIMER/HPET or due to SMI interference then the setup of the per
CPU cyc2ns variables is skipped. This is unlikely to happen but it
would definitely render sched_clock() unusable.
This was introduced with commit 53d517cdba
x86: scale cyc_2_nsec according to CPU frequency
Update the per CPU cyc2ns variables in all exit pathes of tsc_calibrate.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@kernel.org
Unconditionally enable PAT support on Centaur and Transmeta CPUs.
All known models that advertise PAT have no known errata.
Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6:
[PATCH] return to old errno choice in mkdir() et.al.
[Patch] fs/binfmt_elf.c: fix wrong return values
[PATCH] get rid of leak in compat_execve()
[Patch] fs/binfmt_elf.c: fix a wrong free
[PATCH] avoid multiplication overflows and signedness issues for max_fds
[PATCH] dup_fd() part 4 - race fix
[PATCH] dup_fd() - part 3
[PATCH] dup_fd() part 2
[PATCH] dup_fd() fixes, part 1
[PATCH] take init_files to fs/file.c
This patch removes CVS keywords that weren't updated for a long time
from comments.
Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Dave Jones <davej@redhat.com>
The most common error with powernow-k8 is an ACPI _PSS error
caused either by failure to load the ACPI processor module
or a bad parse of the _PSS object. Make the error message
returned to the user in these situations more straightforward
and easier to understand.
-Mark Langsdorf
Operating System Research Center
AMD
Signed-off-by: Mark Langsdorf <mark.langsdorf@amd.com>
Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
Signed-off-by: Dave Jones <davej@redhat.com>
The previous revert of 0c07ee38c9 left
out the mwait disable condition for AMD family 10H/11H CPUs.
Andreas Herrman said:
It depends on the CPU. For AMD CPUs that support MWAIT this is wrong.
Family 0x10 and 0x11 CPUs will enter C1 on HLT. Powersavings then
depend on a clock divisor and current Pstate of the core.
If all cores of a processor are in halt state (C1) the processor can
enter the C1E (C1 enhanced) state. If mwait is used this will never
happen.
Thus HLT saves more power than MWAIT here.
It might be best to switch off the mwait flag for these AMD CPU
families like it was introduced with commit
f039b75471 (x86: Don't use MWAIT on AMD
Family 10)
Re-add the AMD families 10H/11H check and disable the mwait usage for
those.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Vegard Nossum reports:
| powertop shows between 200-400 wakeups/second with the description
| "<kernel IPI>: Rescheduling interrupts" when all processors have load (e.g.
| I need to run two busy-loops on my 2-CPU system for this to show up).
|
| The bisect resulted in this commit:
|
| commit 0c07ee38c9
| Date: Wed Jan 30 13:33:16 2008 +0100
|
| x86: use the correct cpuid method to detect MWAIT support for C states
remove the functional effects of this patch and make mwait unconditional.
A future patch will turn off mwait on specific CPUs where that causes
power to be wasted.
Bisected-by: Vegard Nossum <vegard.nossum@gmail.com>
Tested-by: Vegard Nossum <vegard.nossum@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The user_regset_view table for the 32-bit regsets on the 64-bit build had
the wrong sizes for the FP regsets. This bug had no user-visible effect
(just on kernel modules using the user_regset interfaces and the like).
But the fix is trivial and risk-free.
Signed-off-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Fix this symbol export problem:
Building modules, stage 2.
MODPOST 193 modules
ERROR: "csum_partial" [fs/reiserfs/reiserfs.ko] undefined!
make[1]: *** [__modpost] Error 1
make: *** [modules] Error 2
This is due to a known weakness of symbol exports: if a symbol's
only in-core user is an EXPORT_SYMBOL from a lib-y section, the
symbol is not linked in.
The solution is to move the export to x8664_ksyms_64.c - but the real
solution would be to fix kbuild.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
arch/x86/kernel/setup_64.c:954: warning: passing argument 2 of 'set_bit' from incompatible pointer type
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
After resume on a 2cpu laptop, kernel builds collapse with a sed hang,
sh or make segfault (often on 20295564), real-time signal to cc1 etc.
Several hurdles to jump, but a manually-assisted bisect led to -rc1's
d2bcbad5f3 x86: do not zap_low_mappings
in __smp_prepare_cpus. Though the low mappings were removed at bootup,
they were left behind (with Global flags helping to keep them in TLB)
after resume or cpu online, causing the crashes seen.
Reinstate zap_low_mappings (with local __flush_tlb_all) for each cpu_up
on x86_32. This used to be serialized by smp_commenced_mask: that's now
gone, but a low_mappings flag will do. No need for native_smp_cpus_done
to repeat the zap: let mem_init zap BSP's low mappings just like on UP.
(In passing, fix error code from native_cpu_up: do_boot_cpu returns a
variety of diagnostic values, Dprintk what it says but convert to -EIO.
And save_pg_dir separately before zap_low_mappings: doesn't matter now,
but zapping twice in succession wiped out resume's swsusp_pg_dir.)
That worked well on the duo and one quad, but wouldn't boot 3rd or 4th
cpu on P4 Xeon, oopsing just after unlock_ipi_call_lock. The TLB flush
IPI now being sent reveals a long-standing bug: the booting cpu has its
APIC readied in smp_callin at the top of start_secondary, but isn't put
into the cpu_online_map until just before that unlock_ipi_call_lock.
So native_smp_call_function_mask to online cpus would send_IPI_allbutself,
including the cpu just coming up, though it has been excluded from the
count to wait for: by the time it handles the IPI, the call data on
native_smp_call_function_mask's stack may well have been overwritten.
So fall back to send_IPI_mask while cpu_online_map does not match
cpu_callout_map: perhaps there's a better APICological fix to be
made at the start_secondary end, but I wouldn't know that.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The latest rev of Intel doc AP-485 details a new cache
descriptor that we don't yet support.
A 6MB 24-way assoc L2 cache.
Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cleanup gart handling on amd64 a bit: move common code into
enable_gart_translation , and use symbolic register names where
appropriate.
Signed-off-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
1. use symbolic register names where appropriate.
2. num to bus or slot changing
3. handle for new opteron for bus other than 0
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
because we try to reserve dma32 early, so we have chance to get aperture
from 64M.
with some sequence aperture allocated from RAM, could become E820_RESERVED.
and then if doing a kexec with a big kernel that uncompressed size is above
64M we could have a range conflict with still using gart.
So allocate gart aperture from 512M instead.
Also change the fallback_aper_order to 5, because we don't have chance to get
2G or 4G aperture.
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
some systems are using 32M for gart and agp when memory is less than 4G.
Kernel will reject and try to allcate another 64M that is not needed,
and we will waste 64M of perfectly good RAM.
this patch adds a workaround by checking aper_base/order between NB and
agp bridge. If they are the same, and memory size is less than 4G, it
will allow it.
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
while looking at Rafael J. Wysocki's system boot log,
I found a funny printout:
Node 0: aperture @ de000000 size 32 MB
Aperture too small (32 MB)
AGP bridge at 00:04:00
Aperture from AGP @ de000000 size 4096 MB (APSIZE 0)
Aperture too small (0 MB)
Your BIOS doesn't leave a aperture memory hole
Please enable the IOMMU option in the BIOS setup
This costs you 64 MB of RAM
Mapping aperture over 65536 KB of RAM @ 4000000
...
agpgart: Detected AGP bridge 20
agpgart: Aperture pointing to RAM
agpgart: Aperture from AGP @ de000000 size 4096 MB
agpgart: Aperture too small (0 MB)
agpgart: No usable aperture found.
agpgart: Consider rebooting with iommu=memaper=2 to get a good aperture.
it means BIOS allocated the correct gart on the NB and AGP bridge, but
because a bug in the silicon (the agp bridge reports the wrong order,
it wants 4G instead) the kernel will reject that allocation.
Also, because the size is only 32MB, and we try to get another 64M for gart,
late fix_northbridge can not revert that change because it still reads
the wrong size from agp bridge.
So try to double check the order value from the agp bridge, before calling
aperture_valid().
[ mingo@elte.hu: 32-bit fix. ]
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Move symbolic constants into gart.h, and use them instead of hardcoded
constant.
Signed-off-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Move the 4KSTACKS related code to one place. This allows to un#ifdef
do_IRQ() and share the executed on stack for the stack overflow printk
and the softirq call.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Add KERN_WARNING to the printk as this could not be done in the
original patch, which allegedly only moves code around.
Un#ifdef do_IRQ.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Previously the reporting printk would run on the process stack, which risks
overflow an already low stack. Instead execute it on the interrupt stack.
This makes it more likely for the printk to make it actually out.
It adds one not taken test/branch more to the interrupt path when
stack overflow checking is enabled. We could avoid that by duplicating
more code, but that seemed not worth it.
Based on an observation by Eric Sandeen.
v2: Fix warnings in some configs
Signed-off-by: Andi Kleen <andi@firstfloor.org>
Cc: mingo@elte.hu
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Fixes the build error introduced by my FIRST_SYSTEM_VECTOR patch
Signed-off-by: Alan Mayer <ajm@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
The SGI UV system needs several more system vectors than a vanilla
x86_64 system. Rather than burden the other archs with extra system
vectors that they don't use, change FIRST_SYSTEM_VECTOR to a variable,
so that it can be dynamic.
Signed-off-by: Alan Mayer <ajm@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
The interrupt vector defines are copied 4 times around with minimal
differences. Move them all into asm-x86/irq_vectors.h
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Eliminate the 6 bank restriction in 64 bit mce reporting code. This
restriction is artificial (due to static creation of sysfs files) and 32
bit code does not have any such restriction.
This change helps in reporting the details of machine checks on a
machine check exception with errors in bank 6 and above on CPUs that
support those banks. Without the patch, machine check errors in those
banks are not reported.
We still have 128 (MCE_EXTENDED_BANK) bank restriction instead of max
256 supported in hardware. That is not changed in the patch below as it
will have some user level mcelog utility dependency, with bank 128 being
used for thermal reporting currently.
The patch below does not create sysfs control (bankNctl) for banks
higher than 6 as well. That needs some pre-cleanup in /sysfs mce layout,
removal of per cpu /sysfs entries for bankctl as they are really global
system level control today. That change will follow. This basic change
is critical to report the detailed errors on banks higher than 6.
Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
We have a lot of HPET quirks available which might force enable HPET
even when the BIOS does not enable it. Some of those quirks depend on
the command line option "hpet=force".
Andrew pointed out that hoping that the user will find out about this
boot option is not really helpful.
Emit a kernel info which informs the user about the "hpet=force" boot
option when we enter a quirk which depends on this option and the user
did not provide it.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Add quirk to allow forced usage of HPET on ATI SB400.
I stumbled over machines where HPET is enabled but not reported
by BIOS. This patch configures the HPET base address and makes
it known to the OS.
Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
While reading through the HPET code I realized that the
computation of .mult variables could be done with less
lines of code, resulting in a 1.6% text size saving
for hpet.o
So I propose the following patch, which applies against
today's Linus -git tree.
>From 0c6507e400e9ca5f7f14331e18f8c12baf75a9d3 Mon Sep 17 00:00:00 2001
From: Carlos R. Mafra <crmafra@ift.unesp.br>
Date: Mon, 5 May 2008 19:38:53 -0300
The computation of clocksource_hpet.mult
tmp = (u64)hpet_period << HPET_SHIFT;
do_div(tmp, FSEC_PER_NSEC);
clocksource_hpet.mult = (u32)tmp;
can be streamlined if we note that it is equal to
clocksource_hpet.mult = div_sc(hpet_period, FSEC_PER_NSEC, HPET_SHIFT);
Furthermore, the computation of hpet_clockevent.mult
uint64_t hpet_freq;
hpet_freq = 1000000000000000ULL;
do_div(hpet_freq, hpet_period);
hpet_clockevent.mult = div_sc((unsigned long) hpet_freq,
NSEC_PER_SEC, hpet_clockevent.shift);
can also be streamlined with the observation that hpet_period and hpet_freq are
inverse to each other (in proper units).
So instead of computing hpet_freq and using (schematically)
div_sc(hpet_freq, 10^9, shift) we use the trick of calling with the
arguments in reverse order, div_sc(10^6, hpet_period, shift).
The different power of ten is due to frequency being in Hertz (1/sec)
and the period being in units of femtosecond. Explicitly,
mult = (hpet_freq * 2^shift)/10^9 (before)
mult = (10^6 * 2^shift)/hpet_period (after)
because hpet_freq = 10^15/hpet_period.
The comments in the code are also updated to reflect the changes.
As a result,
text data bss dec hex filename
2957 425 92 3474 d92 arch/x86/kernel/hpet.o
3006 425 92 3523 dc3 arch/x86/kernel/hpet.o.old
a 1.6% reduction in text size.
Signed-off-by: Carlos R. Mafra <crmafra@ift.unesp.br>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Polish the ds.h interface and add support for PEBS.
Ds.c is meant to be the resource allocator for per-thread and per-cpu
BTS and PEBS recording.
It is used by ptrace/utrace to provide execution tracing of debugged tasks.
It will be used by profilers (e.g. perfmon2).
It may be used by kernel debuggers to provide a kernel execution trace.
Changes in detail:
- guard DS and ptrace by CONFIG macros
- separate DS and BTS more clearly
- simplify field accesses
- add functions to manage PEBS buffers
- add simple protection/allocation mechanism
- added support for Atom
Opens:
- buffer overflow handling
Currently, only circular buffers are supported. This is all we need
for debugging. Profilers would want an overflow notification.
This is planned to be added when perfmon2 is made to use the ds.h
interface.
- utrace intermediate layer
Signed-off-by: Markus Metzger <markus.t.metzger@intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
To allow linker to catch sections overlapping we have to declare
them in appropriate order.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
The phys_cpu_present_map is an expected symbol in the SMP harness.
Unfortunately, x86 recently moved this and a few others to
kernel/setup.c where it doesn't quite work because voyager has to
define its own. Use CONFIG_X86_LOCAL_APIC to isolate these
definitions and fix up another area in setup.c where CONFIG_X86_SMP
should be used instead of CONFIG_SMP.
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Cc: toralf.foerster@gmx.de
Cc: Mike Travis <travis@sgi.com>
Cc: Alexey Starikovskiy <astarikovskiy@suse.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Rene Herman reported:
> commit 8779f2fc3b
>
> "x86: don't try to allocate from DMA zone at first"
>
> breaks all of ISA DMA. Or all of ALSA ISA DMA at least. All
> ISA soundcards are silent following that commit -- no error
> messages, everything appears fine, just silence.
That patch is buggy. We had an implicit assumption that
dev = NULL for ISA devices that require 24bit DMA.
The recent work on x86 dma_alloc_coherent() breaks the ISA DMA buffer
allocation, which is represented by "dev = NULL" and requires 24bit
DMA implicitly.
Bisected-by: Rene Herman <rene.herman@keyaccess.nl>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Rene Herman <rene.herman@keyaccess.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6-x86:
x86: rdc: leds build/config fix
x86: sysfs cpu?/topology is empty in 2.6.25 (32-bit Intel system)
x86: revert commit 709f744 ("x86: bitops asm constraint fixes")
x86: restrict keyboard io ports reservation to make ipmi driver work
x86: fix fpu restore from sig return
x86: remove spew print out about bus to node mapping
x86: revert printk format warning change which is for linux-next
x86: cleanup PAT cpu validation
x86: geode: define geode_has_vsa2() even if CONFIG_MGEODE_LX is not set
x86: GEODE: cache results from geode_has_vsa2() and uninline
x86: revert geode config dependency
On some of our (single board computer) boards (x86) we are using an
IPMI controller that uses I/O ports 0x62 and 0x66 for a KCS (keyboard
controller style) IPMI system interface.
Trying to load the openipmi driver fails, because the ports
(0x62/0x66) are reserved for keyboard. keyboard reserves the full
range 0x60-0x6F while it doesn't need to.
Reserve only ports 0x60 and 0x64 for the legacy PS/2 i8042 keyboad
controller instead of 0x60-0x6F to allow the openipmi driver to work.
[ tglx: added 64bit fixup ]
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
