Making a variable page-aligned by using
__attribute__((section(".data.page_aligned"))) is fragile because if
sizeof(variable) is not also a multiple of page size, it leaves
variables in the remainder of the section unaligned.
This patch introduces two new qualifiers, __page_aligned_data and
__page_aligned_bss to set the section *and* the alignment of
variables. This makes page-aligned variables more robust because the
linker will make sure they're aligned properly. Unfortunately it
requires *all* page-aligned data to use these macros...
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch moves the reserve_crashkernel() to setup.c and removes the
architecture-specific version. Both versions were more or less the same.
I tested it on both x86-64 and i386, with CONFIG_KEXEC on and off (so
that it compiles).
Signed-off-by: Bernhard Walle <bwalle@suse.de>
Cc: yhlu.kernel@gmail.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Someone could write 0 bytes to /proc/sgi_uv/ptc_statistics,
causing
optstr[count - 1] = '\0';
to write to who-knows-where.
(Andi Kleen noticed this need from a patch I sent for
similar code in the ia64 world (sn2_ptc_proc_write()).)
(count less than zero is not possible here, as count is unsigned)
Signed-off-by: Cliff Wickman <cpw@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
v6: 6/19 close the security hole in uv_ptc_proc_write())
> Found a potential security hole while doing that:
> static ssize_t uv_ptc_proc_write(struct file *file, const char __user *user,
> size_t count, loff_t *data)
> if (copy_from_user(optstr, user, count))
> return -EFAULT;
>
> is count guaranteed to never be larger than 64?
is fixed below.
It adds tlb_uv.o to the Makefile.
Signed-off-by: Cliff Wickman <cpw@sgi.com>
Cc: mingo@elte.hu
Signed-off-by: Ingo Molnar <mingo@elte.hu>
physid_mask_of_physid() causes a huge stack (12k) to be created if the
number of APICS is large. Replace physid_mask_of_physid() with a
new function that does not create large stacks. This is a problem only
on large x86_64 systems.
this paves the way to increase MAX_APICS.
Signed-off-by: Jack Steiner <steiner@sgi.com>
Cc: linux-mm@kvack.org
Cc: mingo@elte.hu
Cc: tglx@linutronix.de
Signed-off-by: Ingo Molnar <mingo@elte.hu>
fix:
arch/x86/kernel/tlb_uv.c: In function ‘uv_table_bases_init':
arch/x86/kernel/tlb_uv.c:612: error: ‘bau_tabsp' undeclared (first use in this function)
arch/x86/kernel/tlb_uv.c:612: error: (Each undeclared identifier is reported only once
arch/x86/kernel/tlb_uv.c:612: error: for each function it appears in.)
Signed-off-by: Ingo Molnar <mingo@elte.hu>
TLB shootdown for SGI UV.
v1: 6/2 original
v2: 6/3 corrections/improvements per Ingo's review
v3: 6/4 split atomic operations off to a separate patch (Jeremy's review)
v4: 6/12 include <mach_apic.h> rather than <asm/mach-bigsmp/mach_apic.h>
(fixes a !SMP build problem that Ingo found)
fix the index on uv_table_bases[blade]
Signed-off-by: Cliff Wickman <cpw@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
TLB shootdown for SGI UV.
Depends on patch (in tip/x86/irq):
x86-update-macros-used-by-uv-platform.patch Jack Steiner May 29
This patch provides the ability to flush TLB's in cpu's that are not on
the local node. The hardware mechanism for distributing the flush
messages is the UV's "broadcast assist unit".
The hook to intercept TLB shootdown requests is a 2-line change to
native_flush_tlb_others() (arch/x86/kernel/tlb_64.c).
This code has been tested on a hardware simulator. The real hardware
is not yet available.
The shootdown statistics are provided through /proc/sgi_uv/ptc_statistics.
The use of /sys was considered, but would have required the use of
many /sys files. The debugfs was also considered, but these statistics
should be available on an ongoing basis, not just for debugging.
Issues to be fixed later:
- The IRQ for the messaging interrupt is currently hardcoded as 200
(see UV_BAU_MESSAGE). It should be dynamically assigned in the future.
- The use of appropriate udelay()'s is untested, as they are a problem
in the simulator.
Signed-off-by: Cliff Wickman <cpw@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch uses reserve_bootmem_generic() instead of reserve_bootmem()
to reserve the crashkernel memory on x86_64. That's necessary for NUMA
machines, see 00212fef81:
[PATCH] Fix kdump Crash Kernel boot memory reservation for NUMA machines
This patch will fix a boot memory reservation bug that trashes memory on
the ES7000 when loading the kdump crash kernel.
The code in arch/x86_64/kernel/setup.c to reserve boot memory for the crash
kernel uses the non-numa aware "reserve_bootmem" function instead of the
NUMA aware "reserve_bootmem_generic". I checked to make sure that no other
function was using "reserve_bootmem" and found none, except the ones that
had NUMA ifdef'ed out.
I have tested this patch only on an ES7000 with NUMA on and off (numa=off)
in a single (non-NUMA) and multi-cell (NUMA) configurations.
Signed-off-by: Amul Shah <amul.shah@unisys.com>
Looks-good-to: Vivek Goyal <vgoyal@in.ibm.com>
Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The switch-back to reserve_bootmem() was accidentally introduced in
5c3391f9f7 when adding the BOOTMEM_EXCLUSIVE
parameter.
Signed-off-by: Bernhard Walle <bwalle@suse.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch adds a 'flags' parameter to reserve_bootmem_generic() like it
already has been added in reserve_bootmem() with commit
72a7fe3967.
It also changes all users to use BOOTMEM_DEFAULT, which doesn't effectively
change the behaviour. Since the change is x86-specific, I don't think it's
necessary to add a new API for migration. There are only 4 users of that
function.
The change is necessary for the next patch, using reserve_bootmem_generic()
for crashkernel reservation.
Signed-off-by: Bernhard Walle <bwalle@suse.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Fix setup.c printk format warning:
linux-next-20080605/arch/x86/kernel/setup.c: In function 'setup_per_cpu_areas':
linux-next-20080605/arch/x86/kernel/setup.c:173: warning: format '%lu' expects type 'long unsigned int', but argument 2 has type 'ssize_t'
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
sparse mutters:
arch/x86/mm/numa_64.c:195:27: warning: symbol 'end_pfn' shadows an earlier one
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
plat_node_bdata, cmdline, nodemap_addr, nodemap_size are local to
numa_64.c. Make them static
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Without this patch, my link fails with:
arch/x86/kernel/built-in.o(.cpuinit.text+0x3c6e): In function `get_local_pda':
: undefined reference to `_cpu_pda'
arch/x86/kernel/built-in.o(.cpuinit.text+0x3cd1): In function `get_local_pda':
: undefined reference to `after_bootmem'
arch/x86/kernel/built-in.o(.cpuinit.text+0x3cec): In function `get_local_pda':
: undefined reference to `_cpu_pda'
make[2]: *** [.tmp_vmlinux1] Error 1
Caused by commit 766da892634694f795b18b9538407816896fc470
x86: remove static boot_cpu_pda array v2
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
* Remove the boot_cpu_pda array and pointer table from the data section.
Allocate the pointer table and array during init. do_boot_cpu()
will reallocate the pda in node local memory and if the cpu is being
brought up before the bootmem array is released (after_bootmem = 0),
then it will free the initial pda. This will happen for all cpus
present at system startup.
This removes 512k + 32k bytes from the data section.
For inclusion into sched-devel/latest tree.
Based on:
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
+ sched-devel/latest .../mingo/linux-2.6-sched-devel.git
Signed-off-by: Mike Travis <travis@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
* Consolidate node_to_cpumask operations and remove the 256k
byte node_to_cpumask_map. This is done by allocating the
node_to_cpumask_map array after the number of possible nodes
(nr_node_ids) is known.
* Debug printouts when CONFIG_DEBUG_PER_CPU_MAPS is active have
been increased. It now shows faults when calling node_to_cpumask()
and node_to_cpumask_ptr().
For inclusion into sched-devel/latest tree.
Based on:
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
+ sched-devel/latest .../mingo/linux-2.6-sched-devel.git
Signed-off-by: Mike Travis <travis@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
* Restore the nodenumber field in the x86_64 pda. This field is slightly
different than the x86_cpu_to_node_map mainly because it's a static
indication of which node the cpu is on while the cpu to node map is a
dyanamic mapping that may get reset if the cpu goes offline. This also
simplifies the numa_node_id() macro.
For inclusion into sched-devel/latest tree.
Based on:
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
+ sched-devel/latest .../mingo/linux-2.6-sched-devel.git
Signed-off-by: Mike Travis <travis@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
* Introduce a new PER_CPU macro called "EARLY_PER_CPU". This is
used by some per_cpu variables that are initialized and accessed
before there are per_cpu areas allocated.
["Early" in respect to per_cpu variables is "earlier than the per_cpu
areas have been setup".]
This patchset adds these new macros:
DEFINE_EARLY_PER_CPU(_type, _name, _initvalue)
EXPORT_EARLY_PER_CPU_SYMBOL(_name)
DECLARE_EARLY_PER_CPU(_type, _name)
early_per_cpu_ptr(_name)
early_per_cpu_map(_name, _idx)
early_per_cpu(_name, _cpu)
The DEFINE macro defines the per_cpu variable as well as the early
map and pointer. It also initializes the per_cpu variable and map
elements to "_initvalue". The early_* macros provide access to
the initial map (usually setup during system init) and the early
pointer. This pointer is initialized to point to the early map
but is then NULL'ed when the actual per_cpu areas are setup. After
that the per_cpu variable is the correct access to the variable.
The early_per_cpu() macro is not very efficient but does show how to
access the variable if you have a function that can be called both
"early" and "late". It tests the early ptr to be NULL, and if not
then it's still valid. Otherwise, the per_cpu variable is used
instead:
#define early_per_cpu(_name, _cpu) \
(early_per_cpu_ptr(_name) ? \
early_per_cpu_ptr(_name)[_cpu] : \
per_cpu(_name, _cpu))
A better method is to actually check the pointer manually. In the
case below, numa_set_node can be called both "early" and "late":
void __cpuinit numa_set_node(int cpu, int node)
{
int *cpu_to_node_map = early_per_cpu_ptr(x86_cpu_to_node_map);
if (cpu_to_node_map)
cpu_to_node_map[cpu] = node;
else
per_cpu(x86_cpu_to_node_map, cpu) = node;
}
* Add a flag "arch_provides_topology_pointers" that indicates pointers
to topology cpumask_t maps are available. Otherwise, use the function
returning the cpumask_t value. This is useful if cpumask_t set size
is very large to avoid copying data on to/off of the stack.
* The coverage of CONFIG_DEBUG_PER_CPU_MAPS has been increased while
the non-debug case has been optimized a bit.
* Remove an unreferenced compiler warning in drivers/base/topology.c
* Clean up #ifdef in setup.c
For inclusion into sched-devel/latest tree.
Based on:
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
+ sched-devel/latest .../mingo/linux-2.6-sched-devel.git
Signed-off-by: Mike Travis <travis@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
* Increase the limit of NR_CPUS to 4096 and introduce a boolean
called "MAXSMP" which when set (e.g. "allyesconfig"), will set
NR_CPUS = 4096 and NODES_SHIFT = 9 (512).
* Changed max setting for NODES_SHIFT from 15 to 9 to accurately
reflect the real limit.
Signed-off-by: Mike Travis <travis@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Mike Travis wrote:
> Ingo Molnar wrote:
>> * Mike Travis <travis@sgi.com> wrote:
>>
>>> [Ingo - please replace "PATCH 07/11" with this one.]
>>>
>>> * Remove 544k bytes from the kernel by removing the boot_cpu_pda
>>> array from the data section and allocating it during startup.
>>>
>>> Fixed panic in setup_per_cpu_areas when HOTPLUG_CPU not set.
>>>
>>> For inclusion into sched-devel/latest tree.
>> sched-devel.git randconfig testing found another crash with your queue:
>>
>> [ 0.111060] Brought up 1 CPUs
>> [ 0.111986] Total of 1 processors activated (4022.73 BogoMIPS).
>> [ 0.112987] Testing NMI watchdog ... <1>BUG: unable to handle kernel NULL pointer dereference at 0000000000000040
>> [ 0.114982] IP: [<ffffffff8180d4a0>] check_nmi_watchdog+0xb0/0x210
>> [ 0.114982] PGD 0
>> [ 0.114982] Oops: 0000 [1] SMP
>> [ 0.114982] CPU 0
>> [............]
>>
>> http://redhat.com/~mingo/misc/config-Mon_Apr_28_23_25_25_CEST_2008.bad
>> http://redhat.com/~mingo/misc/log-Mon_Apr_28_23_25_25_CEST_2008.bad
>>
>> Ingo
>
> Hi Ingo,
>
> I need a bit more information on your hardware configuration. Building a
> kernel with the above config file started up fine on both the Intel and AMD
> boxes.
>
> Based on the above output it looks like it might be a UP machine?
...
Ok, I think I found it. In check_nmi_watchdog():
for (cpu = 0; cpu < NR_CPUS; cpu++)
prev_nmi_count[cpu] = cpu_pda(cpu)->__nmi_count;
As I mentioned it works fine on both of my systems so could you try it out?
Thanks!
Mike
--
* Change function check_nmi_watchdog() to use nr_cpu_ids instead of NR_CPUS.
Based on:
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
+ sched-devel/latest .../mingo/linux-2.6-sched-devel.git
Signed-off-by: Mike Travis <travis@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
also make bus_numa work even if ACPI_NUMA is not defined.
don't call pxm_to_node again, and use node directly.
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>
a numa system (with multi HT chains) may return node without ram. Aka it
is not online. Try to get an online node, otherwise return -1.
Signed-off-by: Yinghai Lu <yinghai.lu@sun.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Some HP laptops have a problem with their DSDT reporting as
HP/SB400/10000, which includes some code which overrides all temperature
trip points to 16C if the INTIN2 input of the I/O APIC is enabled. This
input is incorrectly designated the ISA IRQ 0 via an interrupt source
override even though it is wired to the output of the master 8259A and
INTIN0 is not connected at all. So far two models have been identified,
namely nx6125 and nx6325.
Use a knob provided by the I/O APIC interrupt registration code to
abandon any attempts to route IRQ 0 through the I/O APIC for these
systems.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Cc: Len Brown <lenb@kernel.org>
Cc: Matthew Garrett <mjg59@srcf.ucam.org>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
As discovered recently some systems exhibit problems when the 8254 timer
IRQ is routed through the I/O APIC. These problems do not affect the
timer IRQ itself and therefore cannot be detected when the correctness of
operation of the interrupt is verified in check_timer(). Therefore the
I/O APIC path of the timer IRQ has to be disabled entirely.
This is a change that lets platforms ask for the timer IRQ not to be
registered in the I/O APIC interrupt tables. The local APIC and ExtINTA
paths are unaffected. This request is only taken into account for ACPI
platforms as MP table systems seem unaffected so far.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Cc: Len Brown <lenb@kernel.org>
Cc: Matthew Garrett <mjg59@srcf.ucam.org>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
"Maciej W. Rozycki" <macro@linux-mips.org> said:
> Given X86_64 selects X86_LOCAL_APIC I am not sure the redundancy seen
>above does not actually obscure the logic behind... I think:
>
> depends on X86_LOCAL_APIC && !X86_VISWS
>
>would be clearer and get the same.
Suggested-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Len Brown <lenb@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
v2: seperate "fix for compiling when MPPARSE is not set" to another patch
make X86_MPPARSE to be selectable only when acpi is set and
X86_MPPARSE will be set if acpi is not set.
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Maciej W. Rozycki <macro@linux-mips.org>
Cc: Len Brown <lenb@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
if the system doesn't have ioapic, we don't need to store entries for mptable
update
also let mp_config_acpi_gsi not call func in mpparse
so later could decouple mpparse with acpi more easily
Reported-by: Daniel Exner <dex@dragonslave.de>
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Daniel Exner <dex@dragonslave.de>
Cc: Len Brown <lenb@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
1. let 64bit support 88 and e801 too
2. introduce default_machine_specific_memory_setup, and reuse it
for voyager
v2: fix 64 bit compiling
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
seperate SRAT finding and parsing from get_memcfg_from_srat,
and let getmemcfg_from_srat only handle array from previous step.
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
we already have summit and etc depends on genericarch,
so use genericarch only.
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
so don't punish all other cpus without that problem when init highmem
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
and make 32-bit resource registration more like 64 bit.
also move probe_roms back to setup_32.c
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Because of the size limits of struct boot_params (zero page), the
maximum number of E820 memory map entries can be passed to kernel is
128. As pointed by Paul Jackson, there is some machine produced by SGI
with so many nodes that the number of E820 memory map entries is more
than 128. To enabling Linux kernel on these system, a new setup data
type named SETUP_E820_EXT is defined to pass additional memory map
entries to Linux kernel.
This patch is based on x86/auto-latest branch of git-x86 tree and has
been tested on x86_64 and i386 platform.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
use early_node_map to init high pages, so we can remove page_is_ram() and
page_is_reserved_early() in the big loop with add_one_highpage
also remove page_is_reserved_early(), it is not needed anymore.
v2: fix the build of other platforms
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
rename update_memory_range to e820_update_range
rename add_memory_region to e820_add_region
to make it more clear that they are about e820 map operations.
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
so we don't get the same value multiple times.
also make mp_config_acpi_legacy_irqs more readable by moving assignments
together.
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Daniel Exner reported IO-APIC enumeration breakage in linux-next.
Alexey Starikovskiy found out that it might be related to
commit 2944e16b25 "x86: update mptable".
use enable_update_mptable to decide if need check before add mp_irqs array.
Reported-by: Daniel Exner <webmaster@dragonslave.de>
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
1. move that before zone_sizes_init ...
2. add free_early for one old one, otherwise it will be be reserved again
when we init highmem.
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
in case we have kva before ramdisk on a node, we still need to use
those ranges.
v2: reserve_early kva ram area, in case there are holes in highmem, to avoid
those area could be treat as free high pages.
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
1. add reserve_bootmem_generic for 32bit
2. change len to unsigned long
3. make early_res_to_bootmem to use it
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
we are checking mptable early for numaq, so don't need to reserve_bootmem
for it. bootmem is not there yet.
do the same thing as 64-bit.
found it on 64g above system from 64-bit kernel kexec to 32 bit kernel with
numaq support.
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
since we now have 32-bit support for e820_register_active_regions(),
we can merge the parsing of the mem=/memmap= boot parameters.
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch uses reserve_bootmem_generic() instead of reserve_bootmem()
to reserve the crashkernel memory on x86_64. That's necessary for NUMA
machines, see 00212fef81:
[PATCH] Fix kdump Crash Kernel boot memory reservation for NUMA machines
This patch will fix a boot memory reservation bug that trashes memory on
the ES7000 when loading the kdump crash kernel.
The code in arch/x86_64/kernel/setup.c to reserve boot memory for the crash
kernel uses the non-numa aware "reserve_bootmem" function instead of the
NUMA aware "reserve_bootmem_generic". I checked to make sure that no other
function was using "reserve_bootmem" and found none, except the ones that
had NUMA ifdef'ed out.
I have tested this patch only on an ES7000 with NUMA on and off (numa=off)
in a single (non-NUMA) and multi-cell (NUMA) configurations.
Signed-off-by: Amul Shah <amul.shah@unisys.com>
Looks-good-to: Vivek Goyal <vgoyal@in.ibm.com>
Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The switch-back to reserve_bootmem() was accidentally introduced in
5c3391f9f7 when adding the BOOTMEM_EXCLUSIVE
parameter.
Signed-off-by: Bernhard Walle <bwalle@suse.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch adds a 'flags' parameter to reserve_bootmem_generic() like it
already has been added in reserve_bootmem() with commit
72a7fe3967.
It also changes all users to use BOOTMEM_DEFAULT, which doesn't effectively
change the behaviour. Since the change is x86-specific, I don't think it's
necessary to add a new API for migration. There are only 4 users of that
function.
The change is necessary for the next patch, using reserve_bootmem_generic()
for crashkernel reservation.
Signed-off-by: Bernhard Walle <bwalle@suse.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
> That helped a lot, the system seems to work normally now.
>
> Here's the relevant snippet from dmesg:
>
> [ 0.108006] ..TIMER: vector=0x30 apic1=0 pin1=2 apic2=-1 pin2=-1
> [ 0.108006] ..MP-BIOS bug: 8254 timer not connected to IO-APIC
> [ 0.108006] ...trying to set up timer (IRQ0) through the 8259A ... <3>
> [ 0.108006] ..... (found apic 0 pin 2) ...<3> failed.
> [ 0.108006] ...trying to set up timer as Virtual Wire IRQ...<3> works.
>
> and the whole thing is at: http://www.sisk.pl/kernel/debug/20080618/dmesg-2.log
Hmm, that only proved the 8259A is indeed wired to the pin #2 of the I/O
APIC.
> I, personally, don't have any and AMD only has SB600 documentation on its
> web page (it's still marked as "AMD confidential" ;-)).
Well, the IC block is most likely the same as that's not rocket science
and once done there is no need to fiddle with that. That written, I am
afraid there is nothing useful about the IC in the document, except that
it's there and consists of an I/O APIC providing 24 inputs and the usual
pair of 8259A cores. Thanks for the reference anyway.
> There is an interrupt controller in there, but I'm not sure if there's any
> 8259A. The northbridge is on the CPU, actually.
I will praise the day someone ships an x86 machine without an 8259A core!
As expressed in another mail I suspect there may actually be a direct
route from the 8254 to INTIN0 in the southbridge -- this is what other
bootstrap logs seen in the Internet suggest. This would mean this
particular BIOS is buggy (is it the latest version?) and provides an
incorrect IRQ override in its ACPI tables, for example because the
responsible block has been blindly copied from a machine using a commoner
wiring. This could be moderately easily fixed up with a quirk based on
the PCI ID (after checking it again, we actually used to have a quirk for
ATI in this area, but the way it was done suggests the issue was not
understood well enough).
Could you please remove the hack sent yesterday and test the patch
provided below? I do hope it builds, but I have no immediate means to
check it. Please report the output. The intent is to test INTIN0
directly before testing INTIN2 through the 8259A. Thanks.
Aside of that, what I have gathered from your reports (please correct me
if I have got it wrong) is that when the through-8259A mode is used, then
after a while 8254 timer interrupts stop arriving. What's interesting,
the "Virtual Wire IRQ" seems to work for you correctly (that's quite an
odd setup where a local APIC input is used in the native mode -- please
post /proc/interrupts for confirmation), which in turn implies the master
8259A drives its INT output as we expect. Why would the I/O APIC input
have problems then? Hmm...
[ mingo@elte.hu: revert the "x86: fix IO APIC breakage on HP nx6325"
version. ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
On Thu, 19 Jun 2008, Rafael J. Wysocki wrote:
> > With such a configuration the "x86: I/O APIC: timer through 8259A
> > second-chance" patch should not matter, because the only change it
> > introduces is an attempt to try the same I/O APIC pin again, but with the
> > IRQ0 line of the master 8259A enabled. That's not a terribly unusual
> > configuration and nothing should get confused in the system.
>
> But it _does_ get confused, really.
Something certainly gets confused, but so far I am not sure which bit
exactly it is, are you?
> > Barring the unlikely possibility of the 8259A actually being wired to
> > INTIN2 of the I/O APIC I can see two possible explanations:
> >
> > 1. The 8259A interrupt actually escapes to the CPU somehow and is handled
> > as an ExtINTA interrupt. This would make the code in check_timer()
> > decide it has found a working configuration, while actually it has been
> > fooled.
[...]
> Here you go:
>
> [ 0.108006] ..TIMER: vector=0x30 apic1=0 pin1=2 apic2=-1 pin2=-1
> [ 0.108006] ..MP-BIOS bug: 8254 timer not connected to IO-APIC
> [ 0.108006] ...trying to set up timer (IRQ0) through the 8259A ... <3>
> [ 0.108006] ..... (found apic 0 pin 2) ...<3> works.
>
> The full dmesg is at: http://www.sisk.pl/kernel/debug/20080618/dmesg-1.log
Thanks. In this case I suspect the case #1 quoted above happens, that is
the 8259A manages to deliver its interrupt somehow. Note at this stage it
is meant to be in the AEOI mode, so it can happily resubmit the interrupt
indefinitely with no additional handling as long as it receives INTA
cycles.
Can you please try the patch below on top of "x86: I/O APIC: timer
through 8259A second-chance" to see whether my hypothesis is true? It
modifies the through-8259A setup path so that the APIC input gets masked,
but the 8259A has the timer interrupt still enabled. Let me know how the
timer interrupt is routed in this case.
Bisected-by: "Rafael J. Wysocki" <rjw@sisk.pl>
Tested-by: "Rafael J. Wysocki" <rjw@sisk.pl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
If configured to use the I/O APIC, the NMI watchdog is deemed to fail if
the chip has been deactivated as a result of "nosmp". Downgrade to the
local APIC watchdog similarly to what is done for the UP case.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
For the UP case the NMI watchdog downgrade is done consistently in
APIC_init_uniprocessor() now. Remove redundant code used only when
BIOS-disabled local APIC is activated.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
If configured to use the I/O APIC, the NMI watchdog is deemed to fail if
the chip will not be used in the UP configuration, because "noapic" has
been specified or the chip is simply not there. Downgrade to the local
APIC watchdog to rectify.
The new #ifdef is ugly, I know. A proper solution is to provide suitable
definitions of smp_found_config, etc. for !CONFIG_X86_IO_APIC in a header.
Likewise the whole if () condition should be moved to a static inline
function. Such clean-ups are beyond the scope of this change and can be
done once the whole issue of the timer has been sorted out.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
nmi_watchdog=1 hangs on 64-bit:
[ 0.250000] Detected 12.564 MHz APIC timer.
[ 0.254178] APIC timer registered as dummy, due to nmi_watchdog=1!
[ 0.260366] Testing NMI watchdog ... <4>WARNING: CPU#0: NMI appears to be stuck (0->0)!
[ ... ]
[ 0.470003] calling genl_init+0x0/0xd0
[ hard hang ]
bisected it down to:
git-bisect start
git-bisect good 1beee8dc8c
git-bisect bad 11582ece0aaa2d0f94f345c08a4ab9997078a083
git-bisect bad 5479c623bb44089844022c03d4c0eb16d5b7a15f
git-bisect bad cfb4c7fabeb499e1c29f9d1878968e37a938e28a
git-bisect good 246dd412d3
git-bisect bad 3f8237eaff7dc1e35fa791dae095574fd974e671
git-bisect good 90e23b13ab849e2a11f00c655eb3a2011b4623be
git-bisect bad 833526a34eeefc117df3191a594c3c3a4f15a9ac
git-bisect good 791b93d3dfaf16c23e978bec0cc0a3dd9d855d63
git-bisect bad 65767c64068f2c93e56a1accfed5c78230ac12d7
git-bisect bad 2abc5c05dd82c188e3bdf6641a274f013348d14b
git-bisect bad 317e1f2597ffb4d4db940577bbe56dc6e881ef07
| 317e1f2597ffb4d4db940577bbe56dc6e881ef07 is first bad commit
| commit 317e1f2597ffb4d4db940577bbe56dc6e881ef07
| Author: Maciej W. Rozycki <macro@linux-mips.org>
| Date: Wed May 21 22:10:22 2008 +0100
| x86: I/O APIC: clean up the 8259A on a NMI watchdog failure
the problem is that in the dummy-lapic branch we rely on the i8259A
but if the NMI watchdog fails we turn off IRQ 0 - which doesnt work
too well ;-)
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Not sure but maybe it is better to use NMI_DISABLED,
will take a look. But for now this patch is not change
anything in logic so it will not hurt/broke the kernel.
For most cases nmi_watchdog assignment is by one of NMI_*
macro so I think there it make sense too.
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
arch/x86/kernel/io_apic_64.c: In function 'check_timer':
arch/x86/kernel/io_apic_64.c:1688: error: 'vector' undeclared (first use in this function)
arch/x86/kernel/io_apic_64.c:1688: error: (Each undeclared identifier is reported only once
arch/x86/kernel/io_apic_64.c:1688: error: for each function it appears in.)
Some systems incorrectly report the ExtINTA pin of the I/O APIC as the
genuine target of the timer interrupt. Here is a change that copies timer
pin information found to the other pin if one has been found only. This
way both a direct and a through-8259A route is tested with the pin letting
these problematic systems work well enough. If no timer pin information
has been found for the I/O APIC, then local APIC variations are tried
only, similarly to what is done without the change (except without the
misleading messages).
Obviously if we try the first-chance path without being told by the BIOS
to do so, we should not complain either, so do not print the message in
this case.
The 64-bit variation should be updated with a call to
replace_pin_at_irq() which can be done with the upcoming merge. Since
add_pin_to_irq() is now always called in the first-chance path, the
condition to require it in the second-chance path no longer happens.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Keep the timer interrupt line masked when reconfiguring its interrupt
redirection entry in the I/O APIC.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Unmask the timer interrupt line set up in the through-8259A mode
explicitly after setup_timer_IRQ0_pin() has set up the I/O APIC interrupt
redirection entry to let the two operations be unbound from each other.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Rename setup_ExtINT_IRQ0_pin() to setup_timer_IRQ0_pin() to better
reflect the upcoming role of a function setting up a (semi-)arbitrary I/O
APIC pin appropriately for the 8254 timer. By "appropriate" the following
settings are meant: edge-triggered, active-high, all the other settings
per-architecture. Adjust comments to reflect code appropriately. No
functional changes.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The LINT0 line of the local APIC is masked in the LVT0 entry in
check_timer() before this function is ever called. Removed the
redundant unmasking for better control.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
For a better control the masking and unmasking of the timer interrupt
line in the 8259A operating in the 'Virtual Wire' mode has been moved out
of setup_ExtINT_IRQ0_pin() now, so remove the redundant calls from the
function.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
When the through-8259A mode is used for the timer, the call to
set_irq_handler() will register a NULL handler name, resulting in
"IO-APIC-<NULL>" reported. Fix by calling ioapic_register_intr() as done
for all the other I/O APIC interrupts.
The 64-bit variation calls set_irq_chip_and_handler_name() here
needlessly and should get fixed with the upcoming merge.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The local APIC interrupt handler gets registered with
set_irq_chip_and_handler_name(), which results in
"local-APIC-edge-fasteoi" reported as the name of the handler. Fix by
removing the type of the handler left over from before the generic
handlers were introduced.
The 64-bit variation should get fixed with the upcoming merge.
NB It should really use the "edge" handler and not the "fasteoi" one,
but that's a separate issue.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
There is no point in keeping the 8259A enabled if the I/O APIC NMI
watchdog has failed and the 8259A is not used to pass through regular
timer interrupts. This fixes problems with some systems where some logic
gets confused.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
If configured to use the I/O APIC, the NMI watchdog is deemed to fail if
the chip has been deactivated as a result of "nosmp". Downgrade to the
local APIC watchdog similarly to what is done for the UP case.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The local APIC is no longer forced off when "nosmp" has been specified.
Correct the message printed.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Disable the 8259A acting in the "virtual wire" mode to keep the interrupt
line inactive while fiddling with local APIC interrupt vector registers
associated with its destination inputs. To be on the safe side,
especially concerning flipping the trigger mode.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Disable the 8259A when routing of the timer interrupt through the chip to
the local APIC of the primary processor has failed.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Remove the "disable_8254_timer" and "enable_8254_timer" kernel
parameters. Now that AEOI acknowledgements are no longer needed for
correct timer operation, the 8259A can be kept disabled unconditionally
unless interrupts, either timer or watchdog ones, are actually passed
through it.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The code that used to be in do_slow_gettimeoffset() that relied on the
IRR bit of the master 8259A PIC for IRQ0 to check the state of the output
timer 0 of the PIT is no longer there. As a result, there is no need to
use the POLL command to acknowledge the timer interrupt in the "8259A
Virtual Wire", except for the NMI watchdog when the i82489DX APIC is used
(this is because this particular APIC treats NMIs as level-triggered and
keeping the input asserted would keep motherboard NMI sources held off for
too long). Remove the unneeded bits and adjust comments accordingly.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Use PAGE_OFFSET macro instead of using 0xffff810000000000UL directly.
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: hpa@zytor.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
1) Remove __meminit from update_pages_count. It is used inside
split_pages()
2) Make the code depend on PROC_FS. Doing statistics for nothing is
useless and not adding useless code is nice to the Linux tiny folks.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Add information about the mapping state of the direct mapping to
/proc/meminfo. I chose /proc/meminfo because that is where all the other
memory statistics are too and it is a generally useful metric even
outside debugging situations. A lot of split kernel pages means the
kernel will run slower.
This way we can see how many large pages are really used for it and how
many are split.
Useful for general insight into the kernel.
v2: Add hotplug locking to 64bit to plug a very obscure theoretical race.
32bit doesn't need it because it doesn't support hotadd for lowmem.
Fix some typos
v3: Rename dpages_cnt
Add CONFIG ifdef for count update as requested by tglx
Expand description
v4: Fix stupid bugs added in v3
Move update_page_count to pageattr.c
Signed-off-by: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
