This patch corrects a number of minor errors in the Ebony device tree:
- Missing (given as 0) cache sizes are added to the CPU node
- device_type properties are removed from nodes which don't
have a reasonably well defined device_type binding. This does require
a very small code change to locate the busses to be probed for
of_platform devices by 'compatible' instead of 'device_type'.
- A node is added for the SRAM controller
- The unit address of the small-flash node is adjusted to
correctly reflect the reg property.
- device_type values for the MAL and ZMII are updated to
reflected more up-to-date versions of the binding.
- An incorrect offset in the partition map for the large-flash
node is corrected.
- Some redundant values, already commented out are removed
entirely.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
At present attempting to build treeImage.initrd.* boot images will
fail, because make will select the treeImage.% rule which also matches
instead of the correct and more specific treeImage.initrd.% rule.
This patch corrects the problem by listing the more specific rule
first.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
ft_set_prop() from flatdevtree.c in the zImage wrapper will either
replace an existing property in the flat device tree, or add a new
property definiion if the given property isn't present.
However, when adding properties, it adds the property definition
immediately before the node's END_NODE tag, potentially after any
subnode definitions for the node. This confuses the kernel flat tree
parser in prom.c which assumes that all property definitions for a
node come before all subnode definitions.
This patch corrects ft_set_prop() so that it adds new properties
before the first subnode, instead of before the END_NODE tag.
Signed-off-by: David Gibson <[email protected]>
Acked-by: Scott Wood <[email protected]>
Acked-by: Mark A. Greer <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
There are no actual implementations of fixup_bigphys_addr() in
arch/powerpc, and with a 64-bit aware ioremap() and so forth, it
should no longer be necessary. This patch removes the last dregs of
fixup_bigphys_addr() from arch/powerpc.
In fact, the only reason this hasn't caused link errors already is
that nobody must have tried using one of the small number of drivers
using io_remap_pfn_range() on one of the small number of platforms
which are 32-bit but define CONFIG_PHYS_64BIT. Nonetheless this fixes
a bug, and should go into 2.6.22.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
drivers/mtd/maps includes flash maps for the Beech and Arctic PowerPC
405LP based boards. However, the 405LP was discontinued before any
quantity were distributed and those boards no longer have kernel
support in general. Therefore, this patch removes this obsolete code.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: David Woodhouse <[email protected]>
Squashes a compiler warning, and provides more useful information in
the case messed up device tree information.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: David Woodhouse <[email protected]>
Commit d1953c8888 removed the use of
4level-fixup.h for 32-bit systems under arch/powerpc. However, I
missed a few things activated on some configurations, resulting in
some warnings (at least with STRICT_MM_TYPECHECKS enabled) and build
errors in some circumstances. This fixes it.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Most architectures defined three macros, MK_IOSPACE_PFN(), GET_IOSPACE()
and GET_PFN() in pgtable.h. However, the only callers of any of these
macros are in Sparc specific code, either in arch/sparc, arch/sparc64 or
drivers/sbus.
This patch removes the redundant macros from all architectures except
sparc and sparc64.
Signed-off-by: David Gibson <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
powerpc_flash_init() implements a broken way of probing for flash
devices supported by the physmap_of driver. It finds all nodes in the
device tree with device_type=="rom" and instantiates of_platform
devices for them. This is fundamentally incompatible with the normal
and correct way of probing for of_platform_bus_probe(). Platforms
which relied on powerpc_flash_init()s behaviour (none are in-tree)
will have to update their platform probing code to correctly probe
busses containing flash devices.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This adds support for early serial debugging via the built in
port on IBM/AMCC PowerPC 44x CPUs. It uses a bolted TLB entry in
address space 1 for the UART's mapping, allowing robust debugging both
before and after the initialization of the MMU.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This adds platform support code for the Ebony (440GP) evaluation
board. This includes both code in arch/powerpc/platforms/44x for
board initialization, and zImage wrapper code to correctly tweak the
flattened device tree based on information from the firmware. The
zImage supports both IBM OpenBIOS (aka "treeboot") and old versions of
uboot which don't support a flattened device tree.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Add a device tree for the Ebony evaluation board (440GP based). This
tree is not complete or finalized. This tree needs a version of dtc
recent enough to include reference-to-labels to process.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This prepares for Ebony/440 support by creating an
arch/powerpc/platforms/44x directory. It is populated with a single
misc_44x.S file, into which is moved the 44x specific reset code from
head_44x.S (on the grounds that we should really stop clogging up the
head_* files with random asm helper routines).
At the same time, we disable the (empty save Kconfig and Makefile)
arch/powerpc/platforms/4xx directory from the arch/powerpc/platforms
Makefile. Contrary to the comment in
arch/powerpc/platforms/4xx/Makefile, attempting to build such an empty
Makefile will fail, thus breaking compile for the 44x platforms we're
about to add. It can go back in once we start porting some of the 40x
platforms (and thus it becomes non-empty).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
For 32-bit systems, powerpc still relies on the 4level-fixup.h hack,
to pretend that the generic pagetable handling stuff is 3-levels
rather than 4. This patch removes this, instead using the newer
pgtable-nopmd.h to handle the elision of both the pud and pmd
pagetable levels (ppc32 pagetables are actually 2 levels).
This removes a little extraneous code, and makes it more easily
compared to the 64-bit pagetable code.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
At present, the serial core always allows setserial in userspace to change the
port address, irq and base clock of any serial port. That makes sense for
legacy ISA ports, but not for (say) embedded ns16550 compatible serial ports
at peculiar addresses. In these cases, the kernel code configuring the ports
must know exactly where they are, and their clocking arrangements (which can
be unusual on embedded boards). It doesn't make sense for userspace to change
these settings.
Therefore, this patch defines a UPF_FIXED_PORT flag for the uart_port
structure. If this flag is set when the serial port is configured, any
attempts to alter the port's type, io address, irq or base clock with
setserial are ignored.
In addition this patch uses the new flag for on-chip serial ports probed in
arch/powerpc/kernel/legacy_serial.c, and for other hard-wired serial ports
probed by drivers/serial/of_serial.c.
Signed-off-by: David Gibson <[email protected]>
Cc: Russell King <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
ppc4xx_sgdma.c is #including asm/dma-mapping.h directly, which should
only ever be included via linux/dma-mapping.h. asm/dma-mapping.h
relies on an enum defined in linux/dma-mapping.h before its own
include. This fixes the problem.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch fixes a couple of missing dependencies in
arch/powerpc/boot/Makefile. First, it ensures that the zlib.h header
is linked in before attempting to build gunzip_util.o, as it is,
building gunzip_util.o usually works, but not always depending on make
order.
Second, it makes the final images which are built using a dts
dependent on that dts, so the image will be correctly rebuilt if the
dts changes. This in turn requires fixing the definition of the dts
variable. CONFIG_DEVICE_TREE from Kconfig will have quotes around it,
which don't matter when passing the variable to a shell, but which
need to be removed when incorporating it into a filename for make's
use.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
32-bit powerpc uses a PTE_FMT macro to handle printk() formatting of
PTE entries (which can vary in type and size). Apparently there was a
good reason for it once, but with current compilers it's simpler just
to workaround the variation with a cast in the printk() itself
(there's only one use).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch makes a few small cleanups to the cuboot code.
- It removes the double layered selection of images, via
cuboot-plat-y, instead having the cuboot platforms directly select a
suitable image-y (this changes the name of the final cuboot image from
plain cuImage to cuImage.<platform>).
- Factors out some code in the wrapper that's potentially
useful to platforms other than uboot.
Signed-off-by: David Gibson <[email protected]>
Acked-by: Scott Wood <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Currently, all 32-bit powerpc platforms use asm-ppc/pgtable.h and
asm-ppc/pgalloc.h, even when otherwise compiled with ARCH=powerpc.
Those asm-ppc files are a fairly nasty tangle of #ifdefs including a
bunch of things which shouldn't be necessary any more in arch/powerpc.
Cleaning up that mess is going to take a while, but this patch is a
first step. It separates the asm-powerpc/pg{alloc,table}.h into 64
bit and 32 bit versions in asm-powerpc, which the basic .h files in
asm-powerpc select based on config. We make a few tiny tweaks to the
innards of the files along the way, making the outermost ifdefs
(double-inclusion protection and __KERNEL__) a little cleaner, and
#including asm-generic/pgtable.h from the top-level
asm-powerpc/pgtable.h (since both the old 32-bit and 64-bit versions
ended with such an #include).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Since we don't have it active by default, the STRICT_MM_TYPECHECKS
option has bitrotted again. This patch fixes a couple of simple build
fixes if the option is selected. First, pud_t mustn't be defined in
page.h on 32-bit systems, because it conflicts with the version in the
generic pud-folding code. Second, pci_32.c is missing a __pgprot()
wrapper call. Third, a couple of PS3 files use constants of type
pgprot_t when they need the raw values, we add pgprot_val() calls to
fix this.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch takes the definitions for the PPC44x MMU (a software loaded
TLB) from asm-ppc/mmu.h, cleans them up of things no longer necessary
in arch/powerpc and puts them in a new asm-powerpc/mmu_44x.h file. It
also substantially simplifies arch/powerpc/mm/44x_mmu.c and makes a
couple of small fixes necessary for the 44x MMU code to build and work
properly in arch/powerpc.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Currently asm-powerpc/mmu.h has definitions for the 64-bit hash based
MMU. If CONFIG_PPC64 is not set, it instead includes asm-ppc/mmu.h
which contains a particularly horrible mess of #ifdefs giving the
definitions for all the various 32-bit MMUs.
It would be nice to have the low level definitions for each MMU type
neatly in their own separate files. It would also be good to wean
arch/powerpc off dependence on the old asm-ppc/mmu.h.
This patch makes a start on such a cleanup by moving the definitions
for the 64-bit hash MMU to their own file, asm-powerpc/mmu_hash64.h.
Definitions for the other MMUs still all come from asm-ppc/mmu.h,
however each MMU type can now be one-by-one moved over to their own
file, in the process cleaning them up stripping them of cruft no
longer necessary in arch/powerpc.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
32-bit powerpc systems define a macro, PHYS_FMT, giving a printf
format string fragment for displaying physical addresses, since most
32-bit powerpc platforms use 32-bit physical addresses but a few use
64-bit physical addresses.
This macro is used in exactly one place, a rare error message, where
we can solve the problem more simply by just unconditionally casting
the address up to 64-bit quantity before formatting it.
This patch does so, meaning that as we bring MMU definitions from
asm-ppc over to asm-powerpc, cleaning them up in the process, we don't
need to implement this ugly macro (which additionally has a very bad
name for something global).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
BenH's commit a741e67969 in powerpc.git,
although (AFAICT) only intended to affect ppc64, also has side-effects
which break 44x. I think 40x, 8xx and Freescale Book E are also
affected, though I haven't tested them.
The problem lies in unconditionally removing flush_tlb_pending() from
the versions of flush_tlb_mm(), flush_tlb_range() and
flush_tlb_kernel_range() used on ppc64 - which are also used the
embedded platforms mentioned above.
The patch below cleans up the convoluted #ifdef logic in tlbflush.h,
in the process restoring the necessary flushes for the software TLB
platforms. There are three sets of definitions for the flushing
hooks: the software TLB versions (revised to avoid using names which
appear to related to TLB batching), the 32-bit hash based versions
(external functions) amd the 64-bit hash based versions (which
implement batching).
It also moves the declaration of update_mmu_cache() to always be in
tlbflush.h (previously it was in tlbflush.h except for PPC64, where it
was in pgtable.h).
Booted on Ebony (440GP) and compiled for 64-bit and 32-bit
multiplatform.
Signed-off-by: David Gibson <[email protected]>
Acked-by: Benjamin Herrenschmidt <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch adds a driver to arch/powerpc/sysdev for the UIC, the
on-chip interrupt controller from IBM/AMCC 4xx chips. It uses the new
irq host mapping infrastructure.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Now that we always take a device tree in arch/powerpc, there's no good
reason not to allow a single kernel to support multiple embedded 4xx
boards - the correct platform code can be selected based on the device
tree information.
Therefore, this patch re-arranges the 4xx Kconfig code to allow this.
In addition we:
- use "select" instead of depends to configure the correct
config options for specific 4xx CPUs and workarounds, which
makes the information about specific boards and CPUs less
scattered.
- Some old, unused (in arch/powerpc) config options are
removed: WANT_EARLY_SERIAL, IBM_OCP, etc.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This cleans up how the zImage code manipulates the kernel
command line. Notable improvements from the old handling:
- Command line manipulation is consolidated into a new
prep_cmdline() function, rather than being scattered across start()
and some helper functions
- Less stack space use: we use just a single global command
line buffer, which can be initialized by an external tool as before,
we no longer need another command line sized buffer on the stack.
- Easier to support platforms whose firmware passes a
commandline, but not a device tree. Platform code can now point new
loader_info fields to the firmware's command line, rather than having
to do early manipulation of the /chosen bootargs property which may
then be rewritten again by the core.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch adds a library of useful device tree manipulation functions
to the zImage library, for use by platform code. These functions are
based on the hooks already in dt_ops, so they're not dependent on a
particular device tree implementation. This patch also slightly
streamlines the code in main.c using these new functions.
This is a consolidation of my work in this area with Scott Wood's
patches to a very similar end.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch adds a reg.h to the zImage code, with common definitions
for accessing system registers. For now, this includes functions for
retrieving the PVR and the stack pointer. This patch then uses the
new reg.h to let start() display the running stack address without
having to explicitly pass the stack as a parameter from the asm code.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch adds the correct attributes to the zImage's versions of
printf to make gcc generate format string mismatch warnings. It also
corrects several minor problems with format strings in the zImage thus
discovered.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch adds documenting comments to the gunzip convenience
functions added in commit ad9d2716cf.
It also removes a stray newline, and an unused global variable.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch updates booting-without-of.txt to describe version 17 of
the flattened device tree format. Version 17 is a small, backwards
compatible change from version 16, adding an extra field giving the
size of the device tree's structure block. At this time, the kernel
has no use for the extra information, however its presence can make
life easier for bootloaders or other software manipulating the tree.
In addition this patch adds information on the size_dt_strings field
of the device tree header, present since version 3 of the flattened
tree format, but omitted from the documentation. It also makes
changes to consistently refer to versions 16 and 17 as versions 16 and
17 in decimal, rather than version 0x10 which was occasionally used
for version 16 previously.
Finally, we also add the new field to the definition of the device
tree header structure in prom.h
Signed-off-by: David Gibson <[email protected]>
Acked-by: Jon Loeliger <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch re-organises the way the zImage wrapper code is entered, to
allow more flexibility on platforms with unusual entry conditions.
After this patch, a platform .o file has two options:
1) It can define a _zimage_start, in which case the platform code gets
control from the very beginning of execution. In this case the
platform code is responsible for relocating the zImage if necessary,
clearing the BSS, performing any platform specific initialization, and
finally calling start() to load and enter the kernel.
2) It can define platform_init(). In this case the generic crt0.S
handles initial entry, and calls platform_init() before calling
start(). The signature of platform_init() is changed, however, to
take up to 5 parameters (in r3..r7) as they come from the platform's
initial loader, instead of a fixed set of parameters based on OF's
usage.
When using the generic crt0.S, the platform .o can optionally
supply a custom stack to use, using the BSS_STACK() macro. If this
is not supplied, the crt0.S will assume that the loader has
supplied a usable stack.
In either case, the platform code communicates information to the
generic code (specifically, a PROM pointer for OF systems, and/or an
initrd image address supplied by the bootloader) via a global
structure "loader_info".
In addition the wrapper script is rearranged to ensure that the
platform .o is always linked first. This means that platforms where
the zImage entry point is at a fixed address or offset, rather than
being encoded in the binary header can be supported using option (1).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch rewrites prep_kernel() in the zImage wrapper code to be
clearer and more flexible. Notable changes:
- Handling of the initrd image from prep_kernel() has moved
into a new prep_initrd() function.
- The address of the initrd image is now added as device tree
properties, as the kernel expects.
- We only copy a packaged initrd image to a new location if it
is in danger of being clobbered when the kernel moves to its final
location, instead of always.
- By default we decompress the kernel directly to address 0,
instead of requiring it to relocate itself. Platforms (such as OF)
where doing this could clobber still-live firmware data structures can
override the vmlinux_alloc hook to provide an alternate place to
decompress the kernel.
- We no longer pass lots of information between functions in
global variables.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
At present, arch/powerpc/boot/main.c includes a gunzip() function
which is a convenient wrapper around zlib. However, it doesn't
conveniently allow decompressing part of an image to one location,
then the remainder to a different address.
This patch adds a new set of more flexible convenience wrappers around
zlib, moving them to their own file, gunzip_util.c, in the process.
These wrappers allow decompressing sections of the compressed image to
different locations. In addition, they transparently handle
uncompressed data, avoiding special case code to handle uncompressed
vmlinux images.
The patch also converts main.c to use the new wrappers, using the new
flexibility to avoid decompressing the vmlinux's ELF header twice as
we did previously. That in turn means we avoid extending our
allocations for the vmlinux to allow space for the extra copy of the
ELF header.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
At present, when an initrd is passed to the kernel used flat device
tree properties, the memory the initrd occupies must also be reserved
in the flat tree's reserve map, or the kernel may overwrite it. That
makes life more complicated than it could be for the bootwrapper.
This patch makes the kernel automatically reserve the initrd's space.
That in turn requires parsing the initrd parameters earlier than they
are currently, in early_init_dt_scan_chosen() instead of
check_for_initrd().
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
At present calling lmb_reserve() (and hence lmb_add_region()) twice
for exactly the same memory region will cause strange behaviour.
This makes life difficult when booting from a flat device tree with
memory reserve map. Which regions are automatically reserved by the
kernel has changed over time, so it's quite possible a newer kernel
could attempt to auto-reserve a region which is also explicitly listed
in the device tree's reserve map, leading to trouble.
This patch avoids the problem by making lmb_reserve() ignore a call to
reserve a previously reserved region. It also removes a now redundant
test designed to avoid one specific case of the problem noted above.
At present, this patch deals only with duplicate reservations of an
identical region. Attempting to reserve two different, but
overlapping regions will still cause problems. I might post another
patch later dealing with this case, but I'm avoiding it now since it
is substantially more complicated to deal with, less likely to occur
and more likely to indicate a genuine bug elsewhere if it does occur.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Getting BenH's new EMAC driver working on 440GP, I found some more
problems in the native mode paths of the new DCR code:
- dcr_map() is supposed to return a dcr_host_t, but the native
version is a macro that doesn't expand to an expression. With native
DCRs, dcr_host_t is an empty structure, so we just use a constructor
expression instead.
- dcr_unmap() uses {} instead of the safer do {} while (0)
idiom to implement a no-op
Here's a fix.
Signed-off-by: David Gibson <[email protected]>
Acked-by: Benjamin Herrenschmidt <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
udbg_early_init() is a function used on 64 bit systems, which
initializes whichever early udbg backend is configured. This function
is not called on 32-bit, however if btext early debug is enabled it
does have an explicit, inline, #ifdef-ed assignment performing
analagous initialization.
This patch makes things more uniform by folding the btext
initialization as an option into udbg_early_init() and calling that
from the 32-bit setup path.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
ARCH=powerpc should not use the ghastly un-multiplatformable tangle of
includes that starts with asm-ppc/ibm4xx.h. This patch removes a
compile-breaking include of it from head_44x.S.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
The new dcr code does not currently compile when configured for native
DCR access on ARCH=powerpc. This patch fixes the problems.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
The powerpc specific version of hugetlb_get_unmapped_area() makes some
unwarranted assumptions about what checks have been made to its
parameters by its callers. This will lead to a BUG_ON() if a 32-bit
process attempts to make a hugepage mapping which extends above
TASK_SIZE (4GB).
I'm not sure if these assumptions came about because they were valid
with earlier versions of the get_unmapped_area() path, or if it was
always broken. Nonetheless this patch fixes the logic, and removes
the crash.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This makes 2 changes to clean up the flat device tree handling
logic in the zImage wrapper.
First, there were two callbacks from the dt_ops structure used for
producing a final flat tree to pass to the kerne: dt_ops.ft_pack()
which packed the flat tree (possibly a no-op) and dt_ops.ft_addr()
which retreived the address of the final blob. Since they were only
ever called together, this patch combines the two into a single new
callback, dt_ops.finalize(). This new callback does whatever
platform-dependent things are necessary to produce a final flat device
tree blob, and returns the blob's addres.
Second, the current logic calls the kernel with a flat device tree if
one is build into the zImage wrapper, otherwise it boots the kernel
with a PROM pointer, expecting the kernel to copy the OF device tree
itself. This approach precludes the possibility of the platform
wrapper code building a flat device tree from whatever
platform-specific information firmware provides. Thus, this patch
takes the more sensible approach of invoking the kernel with a flat
tree if the dt_ops.finalize callback provides one (by whatever means).
So, the dt_ops.finalize callback can be NULL, or can be a function
which returns NULL. In either case, the zImage wrapper logic assumes
that this is a platform with OF and invokes the kernel accordingly.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch makes the handling of the initrd (or initramfs) in the
zImage wrapper a little easier to follow. Instead of passing the
initrd addresses out from prep_kernel() via the cryptic a1 and a2
parameters, use the global struct add_range, 'initrd'. prep_kernel()
already passes information through the 'vmlinux' addr_range struct, so
this seems like a reasonable extension.
Some comments also clarify the logic with prep_kernel(): we use an
initrd included in the zImage if present, otherwise we use an initrd
passed in by the bootloader in the a1 and a2 parameters (yaboot, at
least, uses this mechanism to pass an initrd).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Unlike mmap(), the codepath for brk() creates a vma without first checking
that it doesn't touch a region exclusively reserved for hugepages. On
powerpc, this can allow it to create a normal page vma in a hugepage
region, causing oopses and other badness.
Add a test to prevent this. With this patch, brk() will simply fail if it
attempts to move the break into a hugepage reserved region.
Signed-off-by: David Gibson <[email protected]>
Cc: Adam Litke <[email protected]>
Cc: Hugh Dickins <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Apparently we've copied the todc drivers, for various RTCs used in
embedded machines from ARCH=ppc to ARCH=powerpc, despite the fact that
it's never used in the latter. This patch removes it.
If we ever need these drivers (which we probably shouldn't now the RTC
class stuff is in), we can transfer them one by one from ARCH=ppc,
removing from the hideous abomination which is the todc
"infrastructure".
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Not all architectures have a file named 'defconfig' (e.g. powerpc).
However the make TAGS and make tags targets search such files for tags,
causing an error message when they don't exist. This patch addresses the
problem by instructing xargs not to run the tags program if there are no
matching files.
Signed-off-by: David Gibson <[email protected]>
Cc: Sam Ravnborg <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
At present, ARCH=powerpc kernels can waste considerable space in
pagetables when making large hugepage mappings. Hugepage PTEs go in
PMD pages, but each PMD page maps 256M and so contains only 16
hugepage PTEs (128 bytes of data), but takes up a 1024 byte
allocation. With CONFIG_PPC_64K_PAGES enabled (64k base page size),
the situation is worse. Now hugepage PTEs are at the PTE page level
(also mapping 256M), so we store 16 hugepage PTEs in a 64k allocation.
The PowerPC MMU already means that any 256M region is either all
hugepage, or all normal pages. Thus, with some care, we can use a
different allocation for the hugepage PTE tables and only allocate the
128 bytes necessary.
Signed-off-by: Paul Mackerras <[email protected]>
Quite a long time back, prepare_hugepage_range() replaced
is_aligned_hugepage_range() as the callback from mm/mmap.c to arch code to
verify if an address range is suitable for a hugepage mapping.
is_aligned_hugepage_range() stuck around, but only to implement
prepare_hugepage_range() on archs which didn't implement their own.
Most archs (everything except ia64 and powerpc) used the same
implementation of is_aligned_hugepage_range(). On powerpc, which
implements its own prepare_hugepage_range(), the custom version was never
used.
In addition, "is_aligned_hugepage_range()" was a bad name, because it
suggests it returns true iff the given range is a good hugepage range,
whereas in fact it returns 0-or-error (so the sense is reversed).
This patch cleans up by abolishing is_aligned_hugepage_range(). Instead
prepare_hugepage_range() is defined directly. Most archs use the default
version, which simply checks the given region is aligned to the size of a
hugepage. ia64 and powerpc define custom versions. The ia64 one simply
checks that the range is in the correct address space region in addition to
being suitably aligned. The powerpc version (just as previously) checks
for suitable addresses, and if necessary performs low-level MMU frobbing to
set up new areas for use by hugepages.
No libhugetlbfs testsuite regressions on ppc64 (POWER5 LPAR).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Zhang Yanmin <[email protected]>
Cc: "David S. Miller" <[email protected]>
Cc: Benjamin Herrenschmidt <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: William Lee Irwin III <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
The optional hugepage callback, hugetlb_free_pgd_range() is presently
implemented non-trivially only on ia64 (but I plan to add one for powerpc
shortly). It has its own prototype for the function in asm-ia64/pgtable.h.
However, since the function is called from generic code, it make sense for
its prototype to be in the generic hugetlb.h header file, as the protypes
other arch callbacks already are (prepare_hugepage_range(),
set_huge_pte_at(), etc.). This patch makes it so.
Signed-off-by: David Gibson <[email protected]>
Cc: William Lee Irwin III <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Turns out the hugepage logic in free_pgtables() was doubly broken. The
loop coalescing multiple normal page VMAs into one call to free_pgd_range()
had an off by one error, which could mean it would coalesce one hugepage
VMA into the same bundle (checking 'vma' not 'next' in the loop). I
transferred this bug into the new is_vm_hugetlb_page() based version.
Here's the fix.
This one didn't bite on powerpc previously for the same reason the
is_hugepage_only_range() problem didn't: powerpc's hugetlb_free_pgd_range()
is identical to free_pgd_range(). It didn't bite on ia64 because the
hugepage region is distant enough from any other region that the separated
PMD_SIZE distance test would always prevent coalescing the two together.
No libhugetlbfs testsuite regressions (ppc64, POWER5).
Signed-off-by: David Gibson <[email protected]>
Cc: William Lee Irwin III <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
free_pgtables() has special logic to call hugetlb_free_pgd_range() instead
of the normal free_pgd_range() on hugepage VMAs. However, the test it uses
to do so is incorrect: it calls is_hugepage_only_range on a hugepage sized
range at the start of the vma. is_hugepage_only_range() will return true
if the given range has any intersection with a hugepage address region, and
in this case the given region need not be hugepage aligned. So, for
example, this test can return true if called on, say, a 4k VMA immediately
preceding a (nicely aligned) hugepage VMA.
At present we get away with this because the powerpc version of
hugetlb_free_pgd_range() is just a call to free_pgd_range(). On ia64 (the
only other arch with a non-trivial is_hugepage_only_range()) we get away
with it for a different reason; the hugepage area is not contiguous with
the rest of the user address space, and VMAs are not permitted in between,
so the test can't return a false positive there.
Nonetheless this should be fixed. We do that in the patch below by
replacing the is_hugepage_only_range() test with an explicit test of the
VMA using is_vm_hugetlb_page().
This in turn changes behaviour for platforms where is_hugepage_only_range()
returns false always (everything except powerpc and ia64). We address this
by ensuring that hugetlb_free_pgd_range() is defined to be identical to
free_pgd_range() (instead of a no-op) on everything except ia64. Even so,
it will prevent some otherwise possible coalescing of calls down to
free_pgd_range(). Since this only happens for hugepage VMAs, removing this
small optimization seems unlikely to cause any trouble.
This patch causes no regressions on the libhugetlbfs testsuite - ppc64
POWER5 (8-way), ppc64 G5 (2-way) and i386 Pentium M (UP).
Signed-off-by: David Gibson <[email protected]>
Cc: William Lee Irwin III <[email protected]>
Acked-by: Hugh Dickins <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Originally, mm/hugetlb.c just handled the hugepage physical allocation path
and its {alloc,free}_huge_page() functions were used from the arch specific
hugepage code. These days those functions are only used with mm/hugetlb.c
itself. Therefore, this patch makes them static and removes their
prototypes from hugetlb.h. This requires a small rearrangement of code in
mm/hugetlb.c to avoid a forward declaration.
This patch causes no regressions on the libhugetlbfs testsuite (ppc64,
POWER5).
Signed-off-by: David Gibson <[email protected]>
Cc: William Lee Irwin III <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
These days, hugepages are demand-allocated at first fault time. There's a
somewhat dubious (and racy) heuristic when making a new mmap() to check if
there are enough available hugepages to fully satisfy that mapping.
A particularly obvious case where the heuristic breaks down is where a
process maps its hugepages not as a single chunk, but as a bunch of
individually mmap()ed (or shmat()ed) blocks without touching and
instantiating the pages in between allocations. In this case the size of
each block is compared against the total number of available hugepages.
It's thus easy for the process to become overcommitted, because each block
mapping will succeed, although the total number of hugepages required by
all blocks exceeds the number available. In particular, this defeats such
a program which will detect a mapping failure and adjust its hugepage usage
downward accordingly.
The patch below addresses this problem, by strictly reserving a number of
physical hugepages for hugepage inodes which have been mapped, but not
instatiated. MAP_SHARED mappings are thus "safe" - they will fail on
mmap(), not later with an OOM SIGKILL. MAP_PRIVATE mappings can still
trigger an OOM. (Actually SHARED mappings can technically still OOM, but
only if the sysadmin explicitly reduces the hugepage pool between mapping
and instantiation)
This patch appears to address the problem at hand - it allows DB2 to start
correctly, for instance, which previously suffered the failure described
above.
This patch causes no regressions on the libhugetblfs testsuite, and makes a
test (designed to catch this problem) pass which previously failed (ppc64,
POWER5).
Signed-off-by: David Gibson <[email protected]>
Cc: William Lee Irwin III <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Currently, no lock or mutex is held between allocating a hugepage and
inserting it into the pagetables / page cache. When we do go to insert the
page into pagetables or page cache, we recheck and may free the newly
allocated hugepage. However, since the number of hugepages in the system
is strictly limited, and it's usualy to want to use all of them, this can
still lead to spurious allocation failures.
For example, suppose two processes are both mapping (MAP_SHARED) the same
hugepage file, large enough to consume the entire available hugepage pool.
If they race instantiating the last page in the mapping, they will both
attempt to allocate the last available hugepage. One will fail, of course,
returning OOM from the fault and thus causing the process to be killed,
despite the fact that the entire mapping can, in fact, be instantiated.
The patch fixes this race by the simple method of adding a (sleeping) mutex
to serialize the hugepage fault path between allocation and insertion into
pagetables and/or page cache. It would be possible to avoid the
serialization by catching the allocation failures, waiting on some
condition, then rechecking to see if someone else has instantiated the page
for us. Given the likely frequency of hugepage instantiations, it seems
very doubtful it's worth the extra complexity.
This patch causes no regression on the libhugetlbfs testsuite, and one
test, which can trigger this race now passes where it previously failed.
Actually, the test still sometimes fails, though less often and only as a
shmat() failure, rather processes getting OOM killed by the VM. The dodgy
heuristic tests in fs/hugetlbfs/inode.c for whether there's enough hugepage
space aren't protected by the new mutex, and would be ugly to do so, so
there's still a race there. Another patch to replace those tests with
something saner for this reason as well as others coming...
Signed-off-by: David Gibson <[email protected]>
Cc: William Lee Irwin III <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Move the loops used in mm/hugetlb.c to clear and copy hugepages to their
own functions for clarity. As we do so, we add some checks of need_resched
- we are, after all copying megabytes of memory here. We also add
might_sleep() accordingly. We generally dropped locks around the clear and
copy, already but not everyone has PREEMPT enabled, so we should still be
checking explicitly.
For this to work, we need to remove the clear_huge_page() from
alloc_huge_page(), which is called with the page_table_lock held in the COW
path. We move the clear_huge_page() to just after the alloc_huge_page() in
the hugepage no-page path. In the COW path, the new page is about to be
copied over, so clearing it was just a waste of time anyway. So as a side
effect we also fix the fact that we held the page_table_lock for far too
long in this path by calling alloc_huge_page() under it.
It causes no regressions on the libhugetlbfs testsuite (ppc64, POWER5).
Signed-off-by: David Gibson <[email protected]>
Cc: William Lee Irwin III <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
At present, the powerpc pmd_bad() and pud_bad() macros return false
unless the given pmd or pud is zero. This patch makes these tests
more thorough, checking if the given pmd or pud looks like a plausible
pte page or pmd page pointer respectively. This can result in helpful
error messages when messing with the pagetable code.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
The powerpc pud_ERROR() function misleadingly prints a message
indicating a pmd error. This patch fixes it.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Currently ARCH=powerpc will not compile when STRICT_MM_TYPECHECKS is
turned on and CONFIG_64K_PAGES is turned off. This corrects the
problem.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
One of the parameters to the __pud_free_tlb() macro for powerpc is
incorrect (see patch) . We get away with it by accident, because the one
place the macro is called, the second parameter is a variable named "pud".
Signed-off-by: David Gibson <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: Benjamin Herrenschmidt <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
When hugepages are newly allocated to a file in mm/hugetlb.c, we clear them
with a call to clear_highpage() on each of the subpages. We should be
using clear_user_highpage(): on powerpc, at least, clear_highpage() doesn't
correctly mark the page as icache dirty so if the page is executed shortly
after it's possible to get strange results.
Signed-off-by: David Gibson <[email protected]>
Acked-by: William Lee Irwin III <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
The code to mark a page as icache dirty (so that it will later be
icache-dcache flushed when we try to execute from it) is duplicated in
three places: flush_dcache_page() does this marking and nothing else,
but clear_user_page() and copy_user_page() duplicate it, since those
functions make the page icache dirty themselves.
This patch makes those other functions call flush_dcache_page()
instead, so the logic's all in one place. This will make life less
confusing if we ever need to tweak the details of the the lazy icache
flush mechanism.
arch/powerpc/mm/mem.c | 14 ++------------
1 file changed, 2 insertions(+), 12 deletions(-)
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
At present the lppaca - the structure shared with the iSeries
hypervisor and phyp - is contained within the PACA, our own low-level
per-cpu structure. This doesn't have to be so, the patch below
removes it, making a separate array of lppaca structures.
This saves approximately 500*NR_CPUS bytes of image size and kernel
memory, because we don't need aligning gap between the Linux and
hypervisor portions of every PACA. On the other hand it means an
extra level of dereference in many accesses to the lppaca.
The patch also gets rid of several places where we assign the paca
address to a local variable for no particular reason.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch consolidates the variety of macros used for loading 32 or
64-bit constants in assembler (LOADADDR, LOADBASE, SET_REG_TO_*). The
idea is to make the set of macros consistent across 32 and 64 bit and
to make it more obvious which is the appropriate one to use in a given
situation. The new macros and their semantics are described in the
comments in ppc_asm.h.
In the process, we change several places that were unnecessarily using
immediate loads on ppc64 to use the GOT/TOC. Likewise we cleanup a
couple of places where we were clumsily subtracting PAGE_OFFSET with
asm instructions to use assemble-time arithmetic or the toreal() macro
instead.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
mmap() returns -EINVAL if given a zero length, and thus elf_map() in
binfmt_elf.c does likewise if it attempts to map a (page-aligned) ELF
segment with zero filesize. Such a situation never arises with the default
linker scripts, but there's nothing inherently wrong with zero-filesize
(but non-zero memsize) ELF segments. Custom linker scripts can generate
them, and the kernel should be able to map them; this patch makes it so.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Oops, forgot to compile the VMALLOCBASE/VMALLOC_START patch on
iSeries. VMALLOC_START is defined in pgtable.h whereas previously
VMALLOCBASE was previously defined in page.h. lparmap.c needs to be
updated appropriately.
Booted on iSeries RS64 (now).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
On ppc64, we independently define VMALLOCBASE and VMALLOC_START to be
the same thing: the start of the vmalloc() area at 0xd000000000000000.
VMALLOC_START is used much more widely, including in generic code, so
this patch gets rid of the extraneous VMALLOCBASE.
This does require moving the definitions of region IDs from page_64.h
to pgtable.h, but they don't clearly belong in the former rather than
the latter, anyway. While we're moving them, clean up the definitions
of the REGION_IDs:
- Abolish REGION_SIZE, it was only used once, to define
REGION_MASK anyway
- Define the specific region ids in terms of the REGION_ID()
macro.
- Define KERNEL_REGION_ID in terms of PAGE_OFFSET rather than
KERNELBASE. It amounts to the same thing, but conceptually this is
about the region of the linear mapping (which starts at PAGE_OFFSET)
rather than of the kernel text itself (which is at KERNELBASE).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch removes several unnecessary fields from the paca:
- next_jiffy_update_tb was simply unused. Remove trivially.
- The exdsi exception save area was not used. There were plans to use
it, but they never seem to have gone anywhere. If they ever do, we
can put it back. Remove from the paca, and from asm-offsets.c
- The default_decr field was used from asm, but was only ever assigned
the value of tb_ticks_per_jiffy. Just access tb_ticks_per_jiffy from
asm directly instead.
Built and booted on POWER5 LPAR and iSeries RS64.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
On iSeries, the paca contains, amongst other things an ItLpRegSave
structure used by the hypervisor to save registers. The hypervisor
locates this area through a pointer at the beginning of the paca, so
the structure itself can be located elsewhere. This patch moves the
reg_save area out into its own array. This reduces the amount of
iSeries specific gunk which is visible to general powerpc code via
paca.h
Built and booted on POWER5 LPAR and iSeries RS64.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Currently, the powerpc version of hugetlb_get_unmapped_area() entirely
ignores the hint address. The only way to get a hugepage mapping at a
specified address is with MAP_FIXED, in which case there's no way
(short of parsing /proc/self/maps) for userspace to tell if it will
clobber an existing mapping. This is inconvenient, so the patch below
makes hugepage mappings use the given hint address if possible.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Implement copy-on-write support for hugetlb mappings so MAP_PRIVATE can be
supported. This helps us to safely use hugetlb pages in many more
applications. The patch makes the following changes. If needed, I also have
it broken out according to the following paragraphs.
1. Add a pair of functions to set/clear write access on huge ptes. The
writable check in make_huge_pte is moved out to the caller for use by COW
later.
2. Hugetlb copy-on-write requires special case handling in the following
situations:
- copy_hugetlb_page_range() - Copied pages must be write protected so
a COW fault will be triggered (if necessary) if those pages are written
to.
- find_or_alloc_huge_page() - Only MAP_SHARED pages are added to the
page cache. MAP_PRIVATE pages still need to be locked however.
3. Provide hugetlb_cow() and calls from hugetlb_fault() and
hugetlb_no_page() which handles the COW fault by making the actual copy.
4. Remove the check in hugetlbfs_file_map() so that MAP_PRIVATE mmaps
will be allowed. Make MAP_HUGETLB exempt from the depricated VM_RESERVED
mapping check.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Adam Litke <[email protected]>
Cc: William Lee Irwin III <[email protected]>
Cc: "Seth, Rohit" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
On ppc64, when opening a new hugepage region, we need to make sure any
old normal-page SLBs for the area are flushed on all CPUs. There was
a bug in this logic - after putting the new hugepage area masks into
the thread structure, we copied it into the paca (read by the SLB miss
handler) only on one CPU, not on all. This could cause incorrect SLB
entries to be loaded when a multithreaded program was running
simultaneously on several CPUs. This patch corrects the error,
copying the context information into the PACA on all CPUs using the mm
in question before flushing any existing SLB entries.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
On most powerpc CPUs, the dcache and icache are not coherent so
between writing and executing a page, the caches must be flushed.
Userspace programs assume pages given to them by the kernel are icache
clean, so we must do this flush between the kernel clearing a page and
it being mapped into userspace for execute. We were not doing this
for hugepages, this patch corrects the situation.
We use the same lazy mechanism as we use for normal pages, delaying
the flush until userspace actually attempts to execute from the page
in question.
Tested on G5.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
set_page_dirty() will not cope with being handed a page * which is part of
a compound page, but not the master page in that compound page. This case
can occur via access_process_vm() if you attemp to write to another
process's hugepage memory area using ptrace() (causing an oops or hang).
This patch fixes the bug by only calling set_page_dirty() from
access_process_vm() if the page is not a compound page. We already use a
similar fix in bio_set_pages_dirty() for the case of direct io to
hugepages.
Signed-off-by: David Gibson <[email protected]>
Acked-by: William Irwin <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Blah. The patch [0] I recently sent fixing errors with
in_hugepage_area() and prepare_hugepage_range() for powerpc itself has
an off-by-one bug. Furthermore, the related functions
touches_hugepage_*_range() and within_hugepage_*_range() are also
buggy. Some of the bugs, like those addressed in [0] originated with
commit 7d24f0b8a5 where we tweaked the
semantics of where hugepages are allowed. Other bugs have been there
essentially forever, and are due to the undefined behaviour of '<<'
with shift counts greater than the type width (LOW_ESID_MASK could
return non-zero for high ranges with the right congruences).
The good news is that I now have a testsuite which should pick up
things like this if they creep in again.
[0] "powerpc-fix-for-hugepage-areas-straddling-4gb-boundary"
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Commit 7d24f0b8a5 fixed bugs in the ppc64 SLB
miss handler with respect to hugepage handling, and in the process tweaked
the semantics of the hugepage address masks in mm_context_t.
Unfortunately, it left out a couple of necessary changes to go with that
change. First, the in_hugepage_area() macro was not updated to match,
second prepare_hugepage_range() was not updated to correctly handle
hugepages regions which straddled the 4GB point.
The latter appears only to cause process-hangs when attempting to map such
a region, but the former can cause oopses if a get_user_pages() is
triggered at the wrong point. This patch addresses both bugs.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Currently, if a hugetlbfs is mounted without limits (the default), statfs()
will return -1 for max/free/used blocks. This does not appear to be in
line with normal convention: simple_statfs() and shmem_statfs() both return
0 in similar cases. Worse, it confuses the translation logic in
put_compat_statfs(), causing it to return -EOVERFLOW on such a mount.
This patch alters hugetlbfs_statfs() to return 0 for max/free/used blocks
on a mount without limits. Note that we need the test in the patch below,
rather than just using 0 in the sbinfo structure, because the -1 marked in
the free blocks field is used internally to tell the
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
In fs/compat.c, whenever put_compat_statfs() returns an error, the
containing syscall returns -EFAULT. This is presumably by analogy with the
non-compat case, where any non-zero code from copy_to_user() should be
translated into an EFAULT. However, put_compat_statfs() is also return
-EOVERFLOW. The same applies for put_compat_statfs64().
This bug can be observed with a statfs() on a hugetlbfs directory.
hugetlbfs, when mounted without limits reports available, free and total
blocks as -1 (itself a bug, another patch coming). statfs() will
mysteriously return EFAULT although it's parameters are perfectly valid
addresses.
This patch causes the compat versions of statfs() and statfs64() to
correctly propogate the return values from put_compat_statfs() and
put_compat_statfs64().
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
asm-ppc64/imalloc.h is only included from files in arch/powerpc/mm.
We already have a header for mm local definitions,
arch/powerpc/mm/mmu_decl.h. Thus, this patch moves the contents of
imalloc.h into mmu_decl.h. The only exception are the definitions of
PHBS_IO_BASE, IMALLOC_BASE and IMALLOC_END. Those are moved into
pgtable.h, next to similar definitions of VMALLOC_START and
VMALLOC_SIZE.
Built for multiplatform 32bit and 64bit (ARCH=powerpc).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Since the udbg code in ppc64 has no ppc32 equivalent, move it straight
over into arch/powerpc (and include/asm-powerpc for udbg.h). In time,
we probably want to meld the various bits and pieces of 32-bit early
debugging code into udbg, but for now only include it on
CONFIG_PPC64=y builds. The only change during the move is to
standardise the protecting #ifdef/#define in udbg.h, and move its
banner comment above the initial #ifdef (which seems to be normal
practice).
Built and booted on POWER5 LPAR (ARCH=powerpc and ARCH=ppc64). Built
for 32bit multiplatform (ARCH=powerpc).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch moves a bunch more files from arch/ppc64 and
include/asm-ppc64 which have no equivalents in ppc32 code into
arch/powerpc and include/asm-powerpc. The file affected are:
hvcall.h
proc_ppc64.c
sysfs.c
lparcfg.c
rtas_pci.c
The only changes apart from the move and corresponding Makefile
changes are:
- #ifndef/#define in includes updated to _ASM_POWERPC_ form
- trailing whitespace removed
- comments giving full paths removed
Built and booted on POWER5 LPAR (ARCH=powerpc and ARCH=ppc64), built
for 32-bit powermac (ARCH=powerpc).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch consolidates macros used to generate assembly for
compatibility across different CPUs or configs. A new header,
asm-powerpc/asm-compat.h contains the main compatibility macros. It
uses some preprocessor magic to make the macros suitable both for use
in .S files, and in inline asm in .c files. Headers (bitops.h,
uaccess.h, atomic.h, bug.h) which had their own such compatibility
macros are changed to use asm-compat.h.
ppc_asm.h is now for use in .S files *only*, and a #error enforces
that. As such, we're a lot more careless about namespace pollution
here than in asm-compat.h.
While we're at it, this patch adds a call to the PPC405_ERR77 macro in
futex.h which should have had it already, but didn't.
Built and booted on pSeries, Maple and iSeries (ARCH=powerpc). Built
for 32-bit powermac (ARCH=powerpc) and Walnut (ARCH=ppc).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
scanlog.c is only compiled on pSeries. Thus, this patch moves it to
platforms/pseries.
Built and booted on pSeries LPAR (ARCH=powerpc and ARCH=ppc64). Built
for iSeries (ARCH=powerpc).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
The ppc32 and ppc64 versions of cacheflush.h were almost identical.
The two versions of cache.h are fairly similar, except for a bunch of
register definitions in the ppc32 version which probably belong better
elsewhere. This patch, therefore, merges both headers. Notable
points:
- there are several functions in cacheflush.h which exist only
on ppc32 or only on ppc64. These are handled by #ifdef for now, but
these should probably be consolidated, along with the actual code
behind them later.
- Confusingly, both ppc32 and ppc64 have a
flush_dcache_range(), but they're subtly different: it uses dcbf on
ppc32 and dcbst on ppc64, ppc64 has a flush_inval_dcache_range() which
uses dcbf. These too should be merged and consolidated later.
- Also flush_dcache_range() was defined in cacheflush.h on
ppc64, and in cache.h on ppc32. In the merged version it's in
cacheflush.h
- On ppc32 flush_icache_range() is a normal function from
misc.S. On ppc64, it was wrapper, testing a feature bit before
calling __flush_icache_range() which does the actual flush. This
patch takes the ppc64 approach, which amounts to no change on ppc32,
since CPU_FTR_COHERENT_ICACHE will never be set there, but does mean
renaming flush_icache_range() to __flush_icache_range() in
arch/ppc/kernel/misc.S and arch/powerpc/kernel/misc_32.S
- The PReP register info from asm-ppc/cache.h has moved to
arch/ppc/platforms/prep_setup.c
- The 8xx register info from asm-ppc/cache.h has moved to a
new asm-powerpc/reg_8xx.h, included from reg.h
- flush_dcache_all() was defined on ppc32 (only), but was
never called (although it was exported). Thus this patch removes it
from cacheflush.h and from ARCH=powerpc (misc_32.S) entirely. It's
left in ARCH=ppc for now, with the prototype moved to ppc_ksyms.c.
Built for Walnut (ARCH=ppc), 32-bit multiplatform (pmac, CHRP and PReP
ARCH=ppc, pmac and CHRP ARCH=powerpc). Built and booted on POWER5
LPAR (ARCH=powerpc and ARCH=ppc64).
Built for 32-bit powermac (ARCH=ppc and ARCH=powerpc). Built and
booted on POWER5 LPAR (ARCH=powerpc and ARCH=ppc64). Built and booted
on G5 (ARCH=powerpc)
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch moves a bunch of files from arch/ppc64 and
include/asm-ppc64 which have no equivalents in ppc32 code into
arch/powerpc and include/asm-powerpc. The file affected are:
abs_addr.h
compat.h
lppaca.h
paca.h
tce.h
cpu_setup_power4.S
ioctl32.c
firmware.c
pacaData.c
The only changes apart from the move and corresponding Makefile
changes are:
- #ifndef/#define in includes updated to _ASM_POWERPC_ form
- trailing whitespace removed
- comments giving full paths removed
- pacaData.c renamed paca.c to remove studlyCaps
- Misplaced { moved in lppaca.h
Built and booted on POWER5 LPAR (ARCH=powerpc and ARCH=ppc64), built
for 32-bit powermac (ARCH=powerpc).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch merges current.h. This is a one-big-ifdef merge, but both
versions are so tiny, I think we can live with it. While we're at it,
we get rid of the fairly pointless redirection through get_current()
in the ppc64 version.
Built and booted on POWER5 LPAR (ARCH=powerpc & ARCH=ppc64). Built
for 32-bit pmac (ARCH=powerpc & ARCH=ppc).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Having already merged the ppc and ppc64 versions of signal.c, this
patch finishes the job by merging signal.h. The two versions were
almost identical already. Notable changes:
- We use BITS_PER_LONG to correctly size sigset_t
- Remove some uneeded #includes and struct forward
declarations. This does mean adding an include to signal_32.c which
relied on the indirect inclusion of sigcontext.h
- As the ppc64 version, the merged signal.h has prototypes for
do_signal() and do_signal32(). Thus remove extra prototypes from
ppc_ksyms.c which had them directly.
Built and booted on POWER5 LPAR (ARCH=ppc64 and ARCH=powerpc). Built
for 32-bit powermac (ARCH=ppc and ARCH=powerpc) and Walnut (ARCH=ppc).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
When doing its recursive dependency check, scripts/kconfig/conf uses the flag
SYMBOL_CHECK_DONE to avoid rechecking a symbol it has already checked.
However, that flag is only set at the top level, so if a symbol is first
encountered as a dependency of another symbol it will be rechecked every time
it is encountered until it's encountered at the top level.
This patch adjusts the flag setting so that each symbol will only be checked
once, regardless of whether it is first encountered at the top level, or while
recursing down from another symbol. On complex configurations, this vastly
speeds up scripts/kconfig/conf. The config in the powerpc merge tree is
particularly bad: this patch reduces the time for 'scripts/kconfig/conf -o
arch/powerpc/Kconfig' by a factor of 40 on a G5. That's even including the
time to print the config, so the speedup in the actual checking is more likely
2 or 3 orders of magnitude.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Roman Zippel <[email protected]>
Cc: Sam Ravnborg <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
The code for FUTEX_WAKE_OP calls an arch callback,
futex_atomic_op_inuser(). That callback can return an error code, but
currently the caller assumes any error is EFAULT, and will try various
things to resolve the fault before eventually giving up with EFAULT
(regardless of the original error code). This is not a theoretical case -
arch callbacks currently return -ENOSYS if the opcode they are given is
bogus.
This patch alters the code to detect non-EFAULT errors and return them
directly to the user.
Of course, whether -ENOSYS is the correct return value for the bogus opcode
case, or whether EINVAL would be more appropriate is another question.
Signed-off-by: David Gibson <[email protected]>
Cc: Rusty Russell <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jamie Lokier <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
This patch, however, should be applied on top of the 64k-page-size patch to
fix some problems with hugepage (some pre-existing, another introduced by
this patch).
The patch fixes a bug in the SLB miss handler for hugepages on ppc64
introduced by the dynamic hugepage patch (commit id
c594adad56) due to a misunderstanding of the
srd instruction's behaviour (mea culpa). The problem arises when a 64-bit
process maps some hugepages in the low 4GB of the address space (unusual).
In this case, as well as the 256M segment in question being marked for
hugepages, other segments at 32G intervals will be incorrectly marked for
hugepages.
In the process, this patch tweaks the semantics of the hugepage bitmaps to
be more sensible. Previously, an address below 4G was marked for hugepages
if the appropriate segment bit in the "low areas" bitmask was set *or* if
the low bit in the "high areas" bitmap was set (which would mark all
addresses below 1TB for hugepage). With this patch, any given address is
governed by a single bitmap. Addresses below 4GB are marked for hugepage
if and only if their bit is set in the "low areas" bitmap (256M
granularity). Addresses between 4GB and 1TB are marked for hugepage iff
the low bit in the "high areas" bitmap is set. Higher addresses are marked
for hugepage iff their bit in the "high areas" bitmap is set (1TB
granularity).
To avoid conflicts, this patch must be applied on top of BenH's pending
patch for 64k base page size [0]. As such, this patch also addresses a
hugepage problem introduced by that patch. That patch allows hugepages of
1MB in size on hardware which supports it, however, that won't work when
using 4k pages (4 level pagetable), because in that case hugepage PTEs are
stored at the PMD level, and each PMD entry maps 2MB. This patch simply
disallows hugepages in that case (we can do something cleverer to re-enable
them some other day).
Built, booted, and a handful of hugepage related tests passed on POWER5
LPAR (both ARCH=powerpc and ARCH=ppc64).
[0] http://gate.crashing.org/~benh/ppc64-64k-pages.diff
Signed-off-by: David Gibson <[email protected]>
Cc: Benjamin Herrenschmidt <[email protected]>
Cc: Paul Mackerras <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
The ancient ppcdebug/PPCDBG mechanism is now only used in two places.
First, in the hash setup code, one of the bits allows the size of the
hash table to be reduced by a factor of 8 - which would be better
accomplished with a command line option for that purpose. The other
was a bunch of bus walking related messages in the iSeries code, which
would seem to be insufficient reason to keep the mechanism.
This patch removes the last traces of this mechanism.
Built and booted on iSeries and pSeries POWER5 LPAR (ARCH=powerpc).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Oops, replacing the two u64s in struct ipc64_perm with __u32s changed
the alignment of that structure, which could mess up userspace.
Revert to using two unsigned long longs (which is what ppc32 had
originally). ppc64 orignally had two unsigned longs, but long long is
the same size on 64 bit, so this should be ok there too.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
These days, the NACA only exists on iSeries. Therefore, this patch
moves naca.h from include/asm-ppc64 to arch/powerpc/platforms/iseries.
There was one file including naca.h outside of platforms/iseries -
arch/ppc64/kernel/udbg_scc.c. However, that's obviously a hangover
from older days. The include is not necessary, so this patch simply
removes it.
Built and booted on iSeries, built for G5 (which uses udbg_scc.o).
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
asm-ppc64/dart.h is included in exactly one place -
arch/powerpc/sysdev/u3_iommu.c. This patch, therefore, moves it into
arch/powerpc/sysdev. While we're at it, update the #ifndef/#define
protecting the include, and the filename in the comments of
u3_iommu.c.
Built and booted on pSeries and G5, built for ppc32 powermac.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
The ppc64 versions of numnodes.h and sparsemem.h can be safely moved
to asm-powerpc with no changes apart from changing the #define to the
standard _ASM_POWERPC_ form. There are no ppc32 versions of these
files, because they only have any effect if CONFIG_SPARSEMEM is
enabled, which it never can be on ppc32.
Built and booted on pSeries (POWER5), built for 32-bit powermac.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch merges the ppc32 and ppc64 versions of futex.h, essentially
by taking the ppc64 version as the powerpc version. The old ppc32
version did not implement the futex_atomic_op_inuser() callback (it
always returned -ENOSYS), so FUTEX_WAKE_OP would not work on ppc32.
In fact the ppc64 version of this function is almost suitable for
ppc32 as well - the only change needed is to extend ppc_asm.h with a
macro expanding to to the right pseudo-op to store a pointer (either
".long" or ".llong").
Built and booted on pSeries. Built for 32-bit powermac.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Oops, when merging ipcbuf.h, I forgot that 'u64' can't be used in
user-visible headers. This patch corrects the problem, replacing the
unused fields with an array of four __u32s.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Here's a revised version. This re-introduces the set_bits() function
from ppc64, which I removed because I thought it was unused (it exists
on no other arch). In fact it is used in the powermac interrupt code
(but not on pSeries).
- We use LARXL/STCXL macros to generate the right (32 or 64 bit)
instructions, similar to LDL/STL from ppc_asm.h, used in fpu.S
- ppc32 previously used a full "sync" barrier at the end of
test_and_*_bit(), whereas ppc64 used an "isync". The merged version
uses "isync", since I believe that's sufficient.
- The ppc64 versions of then minix_*() bitmap functions have changed
semantics. Previously on ppc64, these functions were big-endian
(that is bit 0 was the LSB in the first 64-bit, big-endian word).
On ppc32 (and x86, for that matter, they were little-endian. As far
as I can tell, the big-endian usage was simply wrong - I guess
no-one ever tried to use minixfs on ppc64.
- On ppc32 find_next_bit() and find_next_zero_bit() are no longer
inline (they were already out-of-line on ppc64).
- For ppc64, sched_find_first_bit() has moved from mmu_context.h to
the merged bitops. What it was doing in mmu_context.h in the first
place, I have no idea.
- The fls() function is now implemented using the cntlzw instruction
on ppc64, instead of generic_fls(), as it already was on ppc32.
- For ARCH=ppc, this patch requires adding arch/powerpc/lib to the
arch/ppc/Makefile. This in turn requires some changes to
arch/powerpc/lib/Makefile which didn't correctly handle ARCH=ppc.
Built and running on G5.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch merges ppc32 and ppc64 versions of ipcbuf.h. The merge is
essentially trivial, since the structure defined in each version was
already identical. Only wrinkle is that the merged version now
includes linux/types.h in order to get the fixed width integer types.
In fact, the old versions probably should have been including that
anyway, since the file uses various __kernel_*_t types.
Built and booted on G5, built for 32-bit pmac, but not booted, since
the merge tree currently doesn't boot there.
Signed-off-by: David Gibson <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
David Gibson