This adds an 'object-id' file that the spe library can
use to store a pointer to its ELF object. This was
originally meant for use by oprofile, but is now
also used by the GNU debugger, if available.
In order for oprofile to find the location in an spu-elf
binary where an event counter triggered, we need a way
to identify the binary in the first place.
Unfortunately, that binary itself can be embedded in a
powerpc ELF binary. Since we can assume it is mapped into
the effective address space of the running process,
have that one write the pointer value into a new spufs
file.
When a context switch occurs, pass the user value to
the profiler so that can look at the mapped file (with
some care).
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Writing to cntl can be used to stop execution on the
spu and to restart it, reading from cntl gives the
contents of the current status register.
The access is always in ascii, as for most other files.
This was always meant to be there, but we had a little
problem with writing to runctl so it was left out so
far.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Since libspe2 will provide a function that can read/write
multiple mailbox elements at once, the kernel should handle
that efficiently.
read/write on the three mailbox files can now access the
spe context multiple times to operate on any number of
mailbox data elements.
If the spu application keeps writing to its outbound
mailbox, the read call will pick up all the data in a
single system call.
Unfortunately, if the user passes an invalid pointer,
we may lose a mailbox element on read, since we can't
put it back. This probably impossible to solve, if the
user also accesses the mailbox through direct register
access.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This hopefully fixes a long-standing bug in the spu file system.
An spu context comes with local memory that can be either saved
in kernel pages or point directly to a physical SPE.
When mapping the physical SPE, that mapping needs to be cache-inhibited.
For simplicity, we used to map the kernel backing memory that way
too, but unfortunately that was not only inefficient, but also incorrect
because the same page could then be accessed simultaneously through
a cacheable and a cache-inhibited mapping, which is not allowed
by the powerpc specification and in our case caused data inconsistency
for which we did a really ugly workaround in user space.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Add the concept of a gang to spufs as a new type of object.
So far, this has no impact whatsover on scheduling, but makes
it possible to add that later.
A new type of object in spufs is now a spu_gang. It is created
with the spu_create system call with the flags argument set
to SPU_CREATE_GANG (0x2). Inside of a spu_gang, it
is then possible to create spu_context objects, which until
now was only possible at the root of spufs.
There is a new member in struct spu_context pointing to
the spu_gang it belongs to, if any. The spu_gang maintains
a list of spu_context structures that are its children.
This information can then be used in the scheduler in the
future.
There is still a bug that needs to be resolved in this
basic infrastructure regarding the order in which objects
are removed. When the spu_gang file descriptor is closed
before the spu_context descriptors, we leak the dentry
and inode for the gang. Any ideas how to cleanly solve
this are appreciated.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This tries to fix spufs so we have an interface closer to what is
specified in the man page for events returned in the third argument of
spu_run.
Fortunately, libspe has never been using the returned contents of that
register, as they were the same as the return code of spu_run (duh!).
Unlike the specification that we never implemented correctly, we now
require a SPU_CREATE_EVENTS_ENABLED flag passed to spu_create, in
order to get the new behavior. When this flag is not passed, spu_run
will simply ignore the third argument now.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
For better explanation, I break down the page fault handling into steps:
1) There is a page fault caused by DMA operation initiated by SPU and
DMA is suspended.
2) The interrupt handler 'spu_irq_class_1()/__spu_trap_data_map()' is
called and it just wakes up the sleeping spe-manager thread.
3) by PPE scheduler, the corresponding bottom half,
spu_irq_class_1_bottom() is called in process context and DMA is
restarted.
There can be a quite large time gap between 2) and 3) and I found
the following problem:
Between 2) and 3) If the context becomes unbound, 3) is not executed
because when the spe-manager thread is awaken, the context is already
saved. (This situation can happen, for example, when a high priority spe
thread newly started in that time gap)
But the actual problem is that the corresponding SPU context does not
work even if it is bound again to a SPU.
Besides I can see the following warning in mambo simulator when the
context becomes
unbound(in save_mfc_cmd()), i.e. when unbind() is called for the
context after step 2) before 3) :
'WARNING: 61392752237: SPE2: MFC_CMD_QUEUE channel count of 15 is
inconsistent with number of available DMA queue entries of 16'
After I go through available documents, I found that the problem is
because the suspended DMA is not restarted when it is bound again.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch adds NUMA support to the the spufs scheduler.
The new arch/powerpc/platforms/cell/spufs/sched.c is greatly
simplified, in an attempt to reduce complexity while adding
support for NUMA scheduler domains. SPUs are allocated starting
from the calling thread's node, moving to others as supported by
current->cpus_allowed. Preemption is gone as it was buggy, but
should be re-enabled in another patch when stable.
The new arch/powerpc/platforms/cell/spu_base.c maintains idle
lists on a per-node basis, and allows caller to specify which
node(s) an SPU should be allocated from, while passing -1 tells
spu_alloc() that any node is allowed.
Since the patch removes the currently implemented preemptive
scheduling, it is technically a regression, but practically
all users have since migrated to this version, as it is
part of the IBM SDK and the yellowdog distribution, so there
is not much point holding it back while the new preemptive
scheduling patch gets delayed further.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch adds a new "psmap" file to spufs that allows mmap of all of
the problem state mapping of SPEs. It is compatible with 64k pages. In
addition, it removes mmap ability of individual files when using 64k
pages, with the exception of signal1 and signal2 which will both map the
entire 64k page holding both registers. It also removes
CONFIG_SPUFS_MMAP as there is no point in not building mmap support in
spufs.
It goes along a separate patch to libspe implementing usage of that new
file to access problem state registers.
Another patch will follow up to fix races opened up by accessing
the 'runcntl' register directly, which is made possible with this
patch.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This eliminates the i_blksize field from struct inode. Filesystems that want
to provide a per-inode st_blksize can do so by providing their own getattr
routine instead of using the generic_fillattr() function.
Note that some filesystems were providing pretty much random (and incorrect)
values for i_blksize.
[bunk@stusta.de: cleanup]
[akpm@osdl.org: generic_fillattr() fix]
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The following patches reduce the size of the VFS inode structure by 28 bytes
on a UP x86. (It would be more on an x86_64 system). This is a 10% reduction
in the inode size on a UP kernel that is configured in a production mode
(i.e., with no spinlock or other debugging functions enabled; if you want to
save memory taken up by in-core inodes, the first thing you should do is
disable the debugging options; they are responsible for a huge amount of bloat
in the VFS inode structure).
This patch:
The filesystem or device-specific pointer in the inode is inside a union,
which is pretty pointless given that all 30+ users of this field have been
using the void pointer. Get rid of the union and rename it to i_private, with
a comment to explain who is allowed to use the void pointer. This is just a
cleanup, but it allows us to reuse the union 'u' for something something where
the union will actually be used.
[judith@osdl.org: powerpc build fix]
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Signed-off-by: Judith Lebzelter <judith@osdl.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
In the context save/restore code, the SPU MFC command queue purge
code has a bug:
static inline void wait_purge_complete(struct spu_state *csa, struct
spu *spu)
{
struct spu_priv2 __iomem *priv2 = spu->priv2;
/* Save, Step 28:
* Poll MFC_CNTL[Ps] until value '11' is
* read
* (purge complete).
*/
POLL_WHILE_FALSE(in_be64(&priv2->mfc_control_RW)
& MFC_CNTL_PURGE_DMA_COMPLETE);
}
This will exit as soon as _one_ of the 2 bits that compose
MFC_CNTL_PURGE_DMA_COMPLETE is set, and one of them happens to be
"purge in progress"... which means that we'll happily continue
restoring the MFC while it's being purged at the same time.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This fixes a bug where we don't properly map SPE MMIO space as guarded,
causing various test cases to fail, probably due to write combining and other
niceties caused by the lack of the G bit.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
locking init cleanups:
- convert " = SPIN_LOCK_UNLOCKED" to spin_lock_init() or DEFINE_SPINLOCK()
- convert rwlocks in a similar manner
this patch was generated automatically.
Motivation:
- cleanliness
- lockdep needs control of lock initialization, which the open-coded
variants do not give
- it's also useful for -rt and for lock debugging in general
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Extend the get_sb() filesystem operation to take an extra argument that
permits the VFS to pass in the target vfsmount that defines the mountpoint.
The filesystem is then required to manually set the superblock and root dentry
pointers. For most filesystems, this should be done with simple_set_mnt()
which will set the superblock pointer and then set the root dentry to the
superblock's s_root (as per the old default behaviour).
The get_sb() op now returns an integer as there's now no need to return the
superblock pointer.
This patch permits a superblock to be implicitly shared amongst several mount
points, such as can be done with NFS to avoid potential inode aliasing. In
such a case, simple_set_mnt() would not be called, and instead the mnt_root
and mnt_sb would be set directly.
The patch also makes the following changes:
(*) the get_sb_*() convenience functions in the core kernel now take a vfsmount
pointer argument and return an integer, so most filesystems have to change
very little.
(*) If one of the convenience function is not used, then get_sb() should
normally call simple_set_mnt() to instantiate the vfsmount. This will
always return 0, and so can be tail-called from get_sb().
(*) generic_shutdown_super() now calls shrink_dcache_sb() to clean up the
dcache upon superblock destruction rather than shrink_dcache_anon().
This is required because the superblock may now have multiple trees that
aren't actually bound to s_root, but that still need to be cleaned up. The
currently called functions assume that the whole tree is rooted at s_root,
and that anonymous dentries are not the roots of trees which results in
dentries being left unculled.
However, with the way NFS superblock sharing are currently set to be
implemented, these assumptions are violated: the root of the filesystem is
simply a dummy dentry and inode (the real inode for '/' may well be
inaccessible), and all the vfsmounts are rooted on anonymous[*] dentries
with child trees.
[*] Anonymous until discovered from another tree.
(*) The documentation has been adjusted, including the additional bit of
changing ext2_* into foo_* in the documentation.
[akpm@osdl.org: convert ipath_fs, do other stuff]
Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Nathan Scott <nathans@sgi.com>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The SPU context save/restore code is currently built
for a 4k page size and we provide a _shipped version
of it since most people don't have the spu toolchain
that is needed to rebuild that code.
This patch hardcodes the data structures to a 64k
page alignment, which also guarantees 4k alignment
but unfortunately wastes 60k of memory per SPU
context that is created in the running system.
We will follow up on this with another patch to
reduce that overhead or maybe redo the context
save/restore logic to do this part entirely different,
but for now it should make experimental systems
work with either page size.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
At this time, all flags are invalid. Since we are
planning to actually add valid flags in the future,
we better check if any were passed by the user.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch remove 'stop_code' -- discarded member of struct spu.
It is written at initialize and interrupt, but never read
in current implementation.
Signed-off-by: Masato Noguchi <Masato.Noguchi@jp.sony.com>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This changes the hypervisor abstraction of setting cpu affinity to a
higher level to avoid platform dependent interrupt controller
routines. I replaced spu_priv1_ops:spu_int_route_set() with a
new routine spu_priv1_ops:spu_cpu_affinity_set().
As a by-product, this change eliminated what looked like an
existing bug in the set affinity code where spu_int_route_set()
mistakenly called int_stat_get().
Signed-off-by: Geoff Levand <geoffrey.levand@am.sony.com>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
To support muti-platform binaries the spu hypervisor accessor
routines must have runtime binding.
I removed the existing statically linked routines in spu.h
and spu_priv1_mmio.c and created new accessor routines in spu_priv1.h
that operate indirectly through an ops struct spu_priv1_ops.
spu_priv1_mmio.c contains the instance of the accessor routines
for running on raw hardware.
Signed-off-by: Geoff Levand <geoffrey.levand@am.sony.com>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
The save/restore sequence for SPE contexts currently attempts to save
and restore the channel count for SPE channel 1 (the SPU_WriteEventMask
channel. But the CBE architecture (section 9.11.2) clearly states
that this channel does not have an associated count. Hardware simply
ignores the attempt to write this count, but the simulator generates
a warning message.
WARNING: 279721590: SPE7: Attempt to write channel count for CH 1 with
no associated count is ignored.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
The wbox channel count of an spu is now initialized
to four for the saved context. This makes it possible
to write to the mailbox right away without waiting
for the SPE to become scheduled first.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
For performance analysis, it is often interesting to know
which physical SPE a thread is currently running on, and,
more importantly, if it is running at all.
This patch adds a simple attribute to each SPU directory
with that information.
The attribute is read-only and called 'phys-id'. It contains
an ascii string with the number of the physical SPU (e.g.
"0x5"), or alternatively the string "0xffffffff" (32 bit -1)
when it is not running at all at the time that the file
is read.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
spufs currently knows only 4k pages and 16M hugetlb
pages. Make it use the regular methods for deciding on
the SLB bits.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
spufs_rmdir tries to acquire the spufs root
i_mutex, which is already held by spufs_create_thread.
This was tracked as Bug #H9512.
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
A recent change to the way that the mfc file gets mapped made it
impossible to map the SPE Multi-Source Synchronization register
into user space, but that may be needed by some applications.
This restores the missing functionality.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
The spu_base module is rather deeply intermixed with the
core kernel, so it makes sense to have that built-in.
This will let us extend the base in the future without
having to export more core symbols just for it.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Use kzalloc when allocating a new spu context, rather than kmalloc +
zeroing.
Booted & tested on cell.
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
We found that when the 'decrementer' is saved, the PPE saves the current
time 'csa->suspend_time'. When restoring the 'decrementer', (Step 34)
decrementer seems to be adjusted with the number of cycles th= at a spu
thread has not been running.
In that code it is missing a substract ('-') because 'delta_time' is
assigned a not substracted(see bellow).
Acked-by: Mark Nutter <mnutter@us.ibm.com>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Missing include for __NR_syscalls, and missing sys_splice() that
causes build-time failure due to compile-time bounds check on
spu_syscall_table.
Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Mark the f_ops members of inodes as const, as well as fix the
ripple-through this causes by places that copy this f_ops and then "do
stuff" with it.
Signed-off-by: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
spufs_init and spufs_exit should be marked correctly so
they can be removed when not needed.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
the mfc member of a new context was not initialized to zero,
which potentially leads to wild memory accesses.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch is layered on top of CONFIG_SPARSEMEM
and is patterned after direct mapping of LS.
This patch allows mmap() of the following regions:
"mfc", which represents the area from [0x3000 - 0x3fff];
"cntl", which represents the area from [0x4000 - 0x4fff];
"signal1" which begins at offset 0x14000; "signal2" which
begins at offset 0x1c000.
The signal1 & signal2 files may be mmap()'d by regular user
processes. The cntl and mfc file, on the other hand, may
only be accessed if the owning process has CAP_SYS_RAWIO,
because they have the potential to confuse the kernel
with regard to parallel access to the same files with
regular file operations: the kernel always holds a spinlock
when accessing registers in these areas to serialize them,
which can not be guaranteed with user mmaps,
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch adds a new file called 'mfc' to each spufs directory.
The file accepts DMA commands that are a subset of what would
be legal DMA commands for problem state register access. Upon
reading the file, a bitmask is returned with the completed
tag groups set.
The file is meant to be used from an abstraction in libspe
that is added by a different patch.
From the kernel perspective, this means a process can now
offload a memory copy from or into an SPE local store
without having to run code on the SPE itself.
The transfer will only be performed while the SPE is owned
by one thread that is waiting in the spu_run system call
and the data will be transferred into that thread's
address space, independent of which thread started the
transfer.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
An SPU does not have a way to implement system calls
itself, but it can create intercepts to the kernel.
This patch uses the method defined by the JSRE interface
for C99 host library calls from an SPU to implement
Linux system calls. It uses the reserved SPU stop code
0x2104 for this, using the structure layout and syscall
numbers for ppc64-linux.
I'm still undecided wether it is better to have a list
of allowed syscalls or a list of forbidden syscalls,
since we can't allow an SPU to call all syscalls that
are defined for ppc64-linux.
This patch implements the easier choice of them, with a
blacklist that only prevents an SPU from calling anything
that interacts with its own execution, e.g fork, execve,
clone, vfork, exit, spu_run and spu_create and everything
that deals with signals.
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
These symbols are only used in the file that they are defined in,
so they should not be in the global namespace.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
The SPE Book IV indicates that MFC DMA operations must be
suspended and restored on SPU context switch (in Step 8).
This patch adds that operation, which is missing from the
current spufs implementation.
Signed-off-by: Masato Noguchi <Masato.Noguchi@jp.sony.com>
Signed-off-by: Geoff Levand <geoffrey.levand@am.sony.com>
Acked-by: Arnd Bergmann <arndb@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch converts the inode semaphore to a mutex. I have tested it on
XFS and compiled as much as one can consider on an ia64. Anyway your
luck with it might be different.
Modified-by: Ingo Molnar <mingo@elte.hu>
(finished the conversion)
Signed-off-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
For far, all SPU triggered interrupts always end up on
the first SMT thread, which is a bad solution.
This patch implements setting the affinity to the
CPU that was running last when entering execution on
an SPU. This should result in a significant reduction
in IPI calls and better cache locality for SPE thread
specific data.
Signed-off-by: Arnd Bergmann <arndb@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
One local variable is missing an __iomem modifier,
in another place, we pass a completely unused argument
with a missing __user modifier.
Signed-off-by: Arnd Bergmann <arndb@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
In a hypervisor based setup, direct access to the first
priviledged register space can typically not be allowed
to the kernel and has to be implemented through hypervisor
calls.
As suggested by Masato Noguchi, let's abstract the register
access trough a number of function calls. Since there is
currently no public specification of actual hypervisor
calls to implement this, I only provide a place that
makes it easier to hook into.
Cc: Masato Noguchi <Masato.Noguchi@jp.sony.com>
Cc: Geoff Levand <geoff.levand@am.sony.com>
Signed-off-by: Arnd Bergmann <arndb@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
The logic for sys_spu_run keeps growing and it does
not really belong into file.c any more since we
moved away from using regular file operations to our
own syscall.
No functional change in here.
Signed-off-by: Arnd Bergmann <arndb@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Arnd Bergmann