The hugepage arch code provides a number of hook functions/macros
which mirror the functionality of various normal page pte access
functions. Various changes in the normal page accessors (in
particular BenH's recent changes to the handling of lazy icache
flushing and PAGE_EXEC) have caused the hugepage versions to get out
of sync with the originals. In some cases, this is a bug, at least on
some MMU types.
One of the reasons that some hooks were not identical to the normal
page versions, is that the fact we're dealing with a hugepage needed
to be passed down do use the correct dcache-icache flush function.
This patch makes the main flush_dcache_icache_page() function hugepage
aware (by checking for the PageCompound flag). That in turn means we
can make set_huge_pte_at() just a call to set_pte_at() bringing it
back into sync. As a bonus, this lets us remove the
hash_huge_page_do_lazy_icache() function, replacing it with a call to
the hash_page_do_lazy_icache() function it was based on.
Some other hugepage pte access hooks - huge_ptep_get_and_clear() and
huge_ptep_clear_flush() - are not so easily unified, but this patch at
least brings them back into sync with the current versions of the
corresponding normal page functions.
Signed-off-by: David Gibson <dwg@au1.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This patch separates the parts of hugetlbpage.c which are inherently
specific to the hash MMU into a new hugelbpage-hash64.c file.
Signed-off-by: David Gibson <dwg@au1.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This patch simplifies the logic used to initialize hugepages on
powerpc. The somewhat oddly named set_huge_psize() is renamed to
add_huge_page_size() and now does all necessary verification of
whether it's given a valid hugepage sizes (instead of just some) and
instantiates the generic hstate structure (but no more).
hugetlbpage_init() now steps through the available pagesizes, checks
if they're valid for hugepages by calling add_huge_page_size() and
initializes the kmem_caches for the hugepage pagetables. This means
we can now eliminate the mmu_huge_psizes array, since we no longer
need to pass the sizing information for the pagetable caches from
set_huge_psize() into hugetlbpage_init()
Determination of the default huge page size is also moved from the
hash code into the general hugepage code.
Signed-off-by: David Gibson <dwg@au1.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Currently each available hugepage size uses a slightly different
pagetable layout: that is, the bottem level table of pointers to
hugepages is a different size, and may branch off from the normal page
tables at a different level. Every hugepage aware path that needs to
walk the pagetables must therefore look up the hugepage size from the
slice info first, and work out the correct way to walk the pagetables
accordingly. Future hardware is likely to add more possible hugepage
sizes, more layout options and more mess.
This patch, therefore reworks the handling of hugepage pagetables to
reduce this complexity. In the new scheme, instead of having to
consult the slice mask, pagetable walking code can check a flag in the
PGD/PUD/PMD entries to see where to branch off to hugepage pagetables,
and the entry also contains the information (eseentially hugepage
shift) necessary to then interpret that table without recourse to the
slice mask. This scheme can be extended neatly to handle multiple
levels of self-describing "special" hugepage pagetables, although for
now we assume only one level exists.
This approach means that only the pagetable allocation path needs to
know how the pagetables should be set out. All other (hugepage)
pagetable walking paths can just interpret the structure as they go.
There already was a flag bit in PGD/PUD/PMD entries for hugepage
directory pointers, but it was only used for debug. We alter that
flag bit to instead be a 0 in the MSB to indicate a hugepage pagetable
pointer (normally it would be 1 since the pointer lies in the linear
mapping). This means that asm pagetable walking can test for (and
punt on) hugepage pointers with the same test that checks for
unpopulated page directory entries (beq becomes bge), since hugepage
pointers will always be positive, and normal pointers always negative.
While we're at it, we get rid of the confusing (and grep defeating)
#defining of hugepte_shift to be the same thing as mmu_huge_psizes.
Signed-off-by: David Gibson <dwg@au1.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Currently we have a fair bit of rather fiddly code to manage the
various kmem_caches used to store page tables of various levels. We
generally have two caches holding some combination of PGD, PUD and PMD
tables, plus several more for the special hugepage pagetables.
This patch cleans this all up by taking a different approach. Rather
than the caches being designated as for PUDs or for hugeptes for 16M
pages, the caches are simply allocated to be a specific size. Thus
sharing of caches between different types/levels of pagetables happens
naturally. The pagetable size, where needed, is passed around encoded
in the same way as {PGD,PUD,PMD}_INDEX_SIZE; that is n where the
pagetable contains 2^n pointers.
Signed-off-by: David Gibson <dwg@au1.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Currently, hpte_need_flush() only correctly flushes the given address
for normal pages. Callers for hugepages are required to mask the
address themselves.
But hpte_need_flush() already looks up the page sizes for its own
reasons, so this is a rather silly imposition on the callers. This
patch alters it to mask based on the pagesize it has looked up itself,
and removes the awkward masking code in the hugepage caller.
Signed-off-by: David Gibson <dwg@au1.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
When running Active Memory Sharing, the Collaborative Memory Manager (CMM)
may mark some pages as "loaned" with the hypervisor. Periodically, the
CMM will query the hypervisor for a loan request, which is a single signed
value. When kexec'ing into a kdump kernel, the CMM driver in the kdump
kernel is not aware of the pages the previous kernel had marked as "loaned",
so the hypervisor and the CMM driver are out of sync. Fix the CMM driver
to handle this scenario by ignoring requests to decrease the number of loaned
pages if we don't think we have any pages loaned. Pages that are marked as
"loaned" which are not in the balloon will automatically get switched to "active"
the next time we touch the page. This also fixes the case where totalram_pages
is smaller than min_mem_mb, which can occur during kdump.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
get_irq_desc() is a powerpc-specific version of irq_to_desc(). That
is reason enough to remove it, but it also doesn't know about sparse
irq_desc support which irq_to_desc() does (when we enable it).
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Rather than open-coding our own check, use irq_has_action()
to check if an irq has an action - ie. is "in use".
irq_has_action() doesn't take the descriptor lock, but it
shouldn't matter - we're just using it as an indicator
that the irq is in use. disable_irq_nosync() will take
the descriptor lock before doing anything also.
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The irq_desc array consumes quite a lot of space, and for systems
that don't need or can't have 512 irqs it's just wasted space.
The first 16 are reserved for ISA, so the minimum of 32 is really
16 - and no one has asked for more than 512 so leave that as the
maximum.
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The CHRP code has some fishy timer based code to scan the RTAS event
log, which uses a 1KB stack buffer and doesn't even use the results.
The pSeries code as a nicer daemon that allows userspace to read the
event log and basically uses the same RTAS interface
This patch moves rtasd.c out of platform/pseries and makes it usable
by CHRP, after removing the old crufty event log mechanism in there.
The nvram logging part of the daemon is still only available on 64-bit
since the underlying nvram management routines aren't currently shared.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Some of the stuff in /proc/ppc64 such as the RTAS bits are actually
useful to some 32-bit platforms. Rename the file, and create a
symlink on 64-bit for backward compatibility
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Just as with kexec, hibernation may fail even on well-tested platforms:
some PCI device, a driver of which doesn't play well with hibernation,
is enough to break resuming.
Hibernation code is not much platform dependent, and hiding features only
because these were not verified on a particular hardware is
counterproductive: we just prevent the features from being widely tested.
For example, with this patch I just tested hibernation on a MPC83xx
board, and it works quite well, modulo a few drivers that need some
fixing.
So, let's make it possible to select hibernation support for all
PowerPCs, then let's wait for any possible bug reports, and actually fix
(or just collect ;-) the bugs instead of hiding them. If some platforms
really can't stand hibernation, we can make a blacklist, with proper
comments why exactly hibernation doesn't work, whether it is possible to
fix, and what needs to be done to fix it.
CONFIG_HIBERNATION is still =n by default, so the commit doesn't change
anything apart from ability to set it to =y.
I'm not sure if EXPERIMENTAL dependency is needed, I'd rather not add it
for a few reasons:
1) It doesn't matter much, for distro kernels user has no clue that some
feature is experimental. Majority of defconfigs enable EXPERIMENTAL
anyway (90 vs. 4, which, btw, means that EXPERIMENTAL is overused
in Kconfigs);
2) EXPERIMENTAL is a good thing for features that change default
behaviour of a kernel, while for hibernation user has to explicitly
issue 'echo disk > /sys/power/state' to trigger any hibernation bugs;
3) Per init/Kconfig, EXPERIMENTAL is a good thing to scare and discourage
users from 'widespread use of a feature', while we want to encourage
that use.
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The non-debug case in ps3/mm.c uses pr_debug(), so that the compiler
still does type checks etc. and doesn't complain about unused
variables in the non-debug case.
However with DEBUG=n and CONFIG_DYNAMIC_DEBUG=y there's still code
generated for those pr_debugs().
size before:
text data bss dec hex filename
17553 4112 88 21753 54f9 arch/powerpc/platforms/ps3/mm.o
size after:
text data bss dec hex filename
7377 776 88 8241 2031 arch/powerpc/platforms/ps3/mm.o
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Acked-by: Geoff Levand <geoffrey.levand@am.sony.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Based on an original patch by Valentine Barshak <vbarshak@ru.mvista.com>
Use preempt_schedule_irq to prevent infinite irq-entry and
eventual stack overflow problems with fast-paced IRQ sources.
This kind of problems has been observed on the PASemi Electra IDE
controller. We have to make sure we are soft-disabled before calling
preempt_schedule_irq and hard disable interrupts after that
to avoid unrecoverable exceptions.
This patch also moves the "clrrdi r9,r1,THREAD_SHIFT" out of
the #ifdef CONFIG_PPC_BOOK3E scope, since r9 is clobbered
and has to be restored in both cases.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Fix the following 3 issues:
arch/powerpc/kernel/process.c: In function 'arch_randomize_brk':
arch/powerpc/kernel/process.c:1183: error: 'mmu_highuser_ssize' undeclared (first use in this function)
arch/powerpc/kernel/process.c:1183: error: (Each undeclared identifier is reported only once
arch/powerpc/kernel/process.c:1183: error: for each function it appears in.)
arch/powerpc/kernel/process.c:1183: error: 'MMU_SEGSIZE_1T' undeclared (first use in this function)
In file included from arch/powerpc/kernel/setup_64.c:60:
arch/powerpc/include/asm/mmu-hash64.h:132: error: redefinition of 'struct mmu_psize_def'
arch/powerpc/include/asm/mmu-hash64.h:159: error: expected identifier or '(' before numeric constant
arch/powerpc/include/asm/mmu-hash64.h:396: error: conflicting types for 'mm_context_t'
arch/powerpc/include/asm/mmu-book3e.h:184: error: previous declaration of 'mm_context_t' was here
cc1: warnings being treated as errors
arch/powerpc/kernel/pci_64.c: In function 'pcibios_unmap_io_space':
arch/powerpc/kernel/pci_64.c💯 error: unused variable 'res'
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This defconfig's purpose at this time is to help catch compile errors
between Book-3S and Book-3E support in ppc64. It is based on the
ppc64_defconfig with some things disabled that we dont support on
Book-3E right now (hugetlbfs, slices, etc.)
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Prior to the arch/ppc -> arch/powerpc transition, xmon had support for single
stepping on 4xx boards. The functionality was lost when arch/ppc was removed.
This patch restores single step support for 44x boards, and Book-E in general.
Signed-off-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The ABI specifies a 64K alignment, we need to map the vDSO accordingly
Signed-off-by: Andreas Schwab <schwab@linux-m68k.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Due to missing segment assignments the .text section was put in the NOTES
segment (and marked as NOTE section), and the .got was put in the DYNAMIC
segment.
Signed-off-by: Andreas Schwab <schwab@linux-m68k.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The creation of the flattened device tree depended on the compiler
putting the constant strings for an object in a section with a
particular name. This was changed with recent compilers. Do this
explicitly instead.
Without this patch, iseries kernels may silently not boot when built with
some compilers.
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The architecture defines that if MSR PR is set we are in problem state
irrespective of the HV bit. This fixes perf events to reflect this.
Also, on bare metal systems, samples taken in Linux will now be reported
as kernel rather than hypervisor.
Signed-off-by: Michael Neuling <mikey@neuling.org>
CC: paulus@samba.org
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The 'fsl5200-clocking'-property was dropped since
0d1cde2358. Remove all occurences
in dts-files.
Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
- serial Console on PSC1
- 64MB SDRAM
- MTD CFI Flash
- Ethernet FEC
- IDE support
Signed-off-by: Heiko Schocher <hs@denx.de>
Reviewed-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
- serial Console on PSC1
- 64MB SDRAM
- MTD CFI Flash
- Ethernet FEC
- I2C with PCF8563 and Temp. Sensor ADM9240
- IDE support
Signed-off-by: Heiko Schocher <hs@denx.de>
Reviewed-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
making a powerpc target with PCI support, shows the
following warning:
MODPOST vmlinux.o
WARNING: vmlinux.o(.text+0x10430): Section mismatch in reference from the
function pcibios_allocate_bus_resources() to the function .init.text:reparent_resources()
The function pcibios_allocate_bus_resources() references
the function __init reparent_resources().
This is often because pcibios_allocate_bus_resources lacks a __init
annotation or the annotation of reparent_resources is wrong.
This patch fix this warning by removing the __init
annotation before reparent_resources.
Signed-off-by: Heiko Schocher <hs@denx.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Here's a patch that adds the ppc750 CL cpu as supported by oprofile.
Signed-off-by: Dragos Tatulea <dtatulea@ixiacom.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
We need to align before the output section. Having the align inside
the output section causes the linker to put some filler in there,
which makes it a non-empty section, but this section isn't assigned to
a segment so you get a warning from the linker.
Signed-off-by: Sean MacLennan <smaclennan@pikatech.com>
Acked-by: Segher Boessenkool <segher@kernel.crashing.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
'acc' isn't used anywhere and thus triggers gcc warning, which causes
build error with CONFIG_PPC_DISABLE_WERROR=n (default):
cc1: warnings being treated as errors
arch/powerpc/kernel/kgdb.c: In function 'gdb_regs_to_pt_regs':
arch/powerpc/kernel/kgdb.c:289: warning: unused variable 'acc'
make[1]: *** [arch/powerpc/kernel/kgdb.o] Error 1
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Profiling of a page fault scalability microbenchmark shows flush_hash_range
is not calling the batch hpte invalidate hcall (H_BULK_REMOVE).
It turns out we have a duplicate firmware feature for hcall-bulk and the
current setup code stops after finding the first match. This meant we never
batch and always do individual invalidates.
The patch below removes the duplicate and shifts FW_FEATURE_CMO to close
the gap. With the patch applied the single threaded page fault rate improves
from 217169 to 238755 per second on a POWER5 test box, a 10% improvement.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
On pSeries, we always force the IO space to be mapped using 4K
pages even with a 64K base page size to cope with some limitations
in the HV interface to some devices.
However, the SLB miss handler code to discriminate between vmalloc
and ioremap space uses a CPU feature section such that the code
is nop'ed out when the processor support large pages non-cachable
mappings.
Thus, we end up always using the ioremap page size for vmalloc
segments on such processors, causing a discrepency between the
segment and the hash table, and thus a hang continously hashing
the page.
It works for the first segment of the vmalloc space since that
segment is "bolted" in by C code correctly, and thankfully we
almost never use the vmalloc space beyond the first segment,
but the new percpu code made the bug happen.
This fixes it by removing the feature section from the assembly,
we now always do the comparison between vmalloc and ioremap.
Signed-off-by; Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
cppcheck found a memory leak in axon_msi, if dcr_base or dcr_len are zero,
we have already allocated msic, so we should free it in the error path.
Signed-off-by: Eric Sesterhenn <eric.sesterhenn@lsexperts.de>
Acked-by: Michael Ellerman <michael@ellerman.id.au>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Since the change of how interrupts are disabled during suspend,
certain PowerBook models started exhibiting various issues during
suspend or resume from sleep.
I finally tracked it down to the code that runs various "platform"
functions (kind of little scripts extracted from the device-tree),
which uses our i2c and PMU drivers expecting interrutps to work,
and at a time where with the new scheme, they have been disabled.
This causes timeouts internally which for some reason results in
the PMU being unable to see the trackpad, among other issues, really
it depends on the machine. Most of the time, we fail to properly adjust
some clocks for suspend/resume so the results are not always
predictable.
This patch fixes it by using IRQF_TIMER for both the PMU and the I2C
interrupts. I prefer doing it this way than moving the call sites since
I really want those platform functions to still be called after all
drivers (and before sysdevs).
We also do a slight cleanup to via-pmu.c driver to make sure the
ADB autopoll mask is handled correctly when doing bus resets
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The mod_return_to_handler needs to switch to the kernel TOC before
jumping to a the kernel code. It currently does this by looking
at the kernel function data and retrieves the TOC that way.
Not only is this inefficient, it also breaks with a relocatable kernel.
The PACA contains the kernel TOC and we can easily retrieve it that
way.
Reported-by: Sachin Sant <sachinp@in.ibm.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
When the function graph tracer is enabled, it replaces the return address
with a hook back to the tracer. This makes back traces see the hook instead
of the actual return address.
The current code also shows the real address by checking if the return
address jumps to the return_to_handler. If it is, is also prints out
the saved real return address.
On powerpc64, some modules may return to mod_return_to_handler, which
is not checked. This patch will also show the real address if a return
is to mod_return_to_handler as well.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* mark struct vm_area_struct::vm_ops as const
* mark vm_ops in AGP code
But leave TTM code alone, something is fishy there with global vm_ops
being used.
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc:
Fix build of cpm_uart due to core changes
powerpc/8xx: Fix regression introduced by cache coherency rewrite
powerpc/4xx: Fix erroneous xmon warning on PowerPC 4xx
powerpc/mm: Fix 40x and 8xx vs. _PAGE_SPECIAL
powerpc: Cleanup linker script using new linker script macros.
powerpc: Fix ibm,client-architecture-support printout
powerpc: Increase NODES_SHIFT on 64bit from 4 to 8
powerpc/perf_counter: Fix vdso detection
powerpc: Move 64bit heap above 1TB on machines with 1TB segments
powerpc: Change archdata dma_data to a union
powerpc: Rename get_dma_direct_offset get_dma_offset
powerpc/mm: Remove duplicated #include
powerpc/book3e-64: Remove duplicated #include
powerpc: Check for unsupported relocs when using CONFIG_RELOCATABLE
powerpc/pmc: Don't access lppaca on Book3E
powerpc: kmalloc failure ignored in vio_build_iommu_table()
hvc_console: Provide (un)locked version for hvc_resize()
* 'for-linus' of git://neil.brown.name/md: (97 commits)
md: raid-1/10: fix RW bits manipulation
md: remove unnecessary memset from multipath.
md: report device as congested when suspended
md: Improve name of threads created by md_register_thread
md: remove sparse warnings about lock context.
md: remove sparse waring "symbol xxx shadows an earlier one"
async_tx/raid6: add missing dma_unmap calls to the async fail case
ioat3: fix uninitialized var warnings
drivers/dma/ioat/dma_v2.c: fix warnings
raid6test: fix stack overflow
ioat2: clarify ring size limits
md/raid6: cleanup ops_run_compute6_2
md/raid6: eliminate BUG_ON with side effect
dca: module load should not be an error message
ioat: driver version 4.0
dca: registering requesters in multiple dca domains
async_tx: remove HIGHMEM64G restriction
dmaengine: sh: Add Support SuperH DMA Engine driver
dmaengine: Move all map_sg/unmap_sg for slave channel to its client
fsldma: Add DMA_SLAVE support
...
After upgrading to the latest kernel on my mpc875 userspace started
running incredibly slow (hours to get to a shell, even!).
I tracked it down to commit 8d30c14cab,
that patch removed a work-around for the 8xx. Adding it
back makes my problem go away.
Signed-off-by: Rex Feany <rfeany@mrv.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The xmon code relies on MSR_RI being non-zero to indicate that an exception
is recoverable. If it is not, it prints a warning message. However, the
PowerPC 4xx cores do not have an MSR_RI bit and this warning is produced for
every xmon event.
This introduces an unrecoverable_excp function to determine if an exception
is recoverable or not. This gets rid of the erroneous warnings on 4xx.
Signed-off-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The test to check whether we have _PAGE_SPECIAL defined is broken,
since we always define it, just not always to a meaninful value :-)
That broke 8xx and 40x under some circumstances.
This fixes it by adding _PAGE_SPECIAL for both of these since they
had a free PTE bit, and removing the condition around advertising
it.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Tim Abbott <tabbott@ksplice.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: linuxppc-dev@ozlabs.org
Cc: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
On machines without the ibm,client-architecture-support call we were missing a
newline. We may as well print the full name in all its glory too - its
ibm,client-architecture-support, not ibm,client-architecture as I mistakenly
wrote (a name only an IBM architect could love).
For my penance I will write out ibm,client-architecture-support 100 times.
Before:
Calling ibm,client-architecture...command line: root=/dev/sda6 console=hvc0 quiet
After:
Calling ibm,client-architecture-support... not implemented
command line: root=/dev/sda6 console=hvc0
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Some System p configurations can already have more than 16 nodes so we
need to increase NODES_SHIFT. I chose 256 to give us some room to grow in the
future, although we can look at something smaller if the memory bloat is
considered too much.
Unless we clamp MAX_ACTIVE_REGIONS we end up with 300kB of extra bloat in
early_node_map in mm/page_alloc.c:
< 6144 early_node_map
> 307200 early_node_map
due to:
#if MAX_NUMNODES >= 32
/* If there can be many nodes, allow up to 50 holes per node */
#define MAX_ACTIVE_REGIONS (MAX_NUMNODES*50)
#else
/* By default, allow up to 256 distinct regions */
#define MAX_ACTIVE_REGIONS 256
Since our memory is mostly contiguous it seems reasonable to keep this
at 256 for now. I also set 32bit to 32 to save space (is there any chance
a 32bit system will have more than 32 discontiguous memory ranges?).
Even with that fixed we have a few data structures that grow:
< 896 bootmem_node_data
> 14336 bootmem_node_data
< 1280 node_devices
> 20480 node_devices
< 25088 kmalloc_caches
> 59648 kmalloc_caches
< 1632 hstates
> 21792 hstates
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
