Commit graph

111 commits

Author SHA1 Message Date
Nick Piggin
7835e98b2e [PATCH] remove set_page_count() outside mm/
set_page_count usage outside mm/ is limited to setting the refcount to 1.
Remove set_page_count from outside mm/, and replace those users with
init_page_count() and set_page_refcounted().

This allows more debug checking, and tighter control on how code is allowed
to play around with page->_count.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-03-22 07:54:02 -08:00
Ingo Molnar
06027bdd27 [PATCH] hrtimer: round up relative start time on low-res arches
CONFIG_TIME_LOW_RES is a temporary way for architectures to signal that
they simply return xtime in do_gettimeoffset().  In this corner-case we
want to round up by resolution when starting a relative timer, to avoid
short timeouts.  This will go away with the GTOD framework.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Cc: Roman Zippel <zippel@linux-m68k.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-02-14 16:09:35 -08:00
Kyle McMartin
108dff7d2b [PATCH] sys_newfstatat -> sys_fstatat64
parisc defines ARCH_WANT_STAT64, so we want to use fstatat64. It does not
appear that it needs to be ENTRY_COMP, because struct stat64 is the same
on both 32-bit and 64-bit (unlike on other platforms which did define a
compat_sys_fstatat64.)

Signed-off-by: Grant Grundler <grundler@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-02-13 19:58:47 -08:00
Kyle McMartin
28b2425ac7 [PARISC] Stub out pselect6/ppoll until TIF_RESTORE_SIGMASK is done
Swap out pselect6/ppoll for ni_syscall for now. We also have to switch
the macro to ENTRY_SAME since compat_sys_ni_syscall does not exist.

Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-02-08 23:27:58 -05:00
Kyle McMartin
1fcbf053e5 [PATCH] sys_hpux: fix strlen_user() race
Userspace can alter the string after the kernel has run strlen_user().

Also: the strlen_user() return value includes the \0, so fix that.

Also: handle EFAULT from strlen_user().

It's unlikely anyone is using this code.  Very, very unlikely.  If I
remember correctly, CONFIG_HPUX turns this code on, but one would actually
need CONFIG_BINFMT_SOM to load a binary that could cause a problem, and
BINFMT_SOM has had an #error in it for quite some time.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-02-07 16:12:33 -08:00
Kyle McMartin
68ee3eb829 [PARISC] New syscalls (inotify, *at, pselect6/ppoll, migrate_pages)
Wire up some new syscalls that have been merged upstream,
  o inotify
  o openat et al
  o pselect6/ppoll
  o migrate_pages

Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-30 00:53:14 -05:00
Kyle McMartin
a2ba82cd7d [PARISC] Simplify DISCONTIGMEM in Kconfig
parisc was previously displaying both the mm/Kconfig memory model
menu, and its own prompt. Remove prompt and have CONFIG_64BIT toggle
between DISCONTIGMEM and FLATMEM. Also remove the EXPERIMENTAL from
discontigmem support... It's been running fine for months (years?)
now.

Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-29 19:33:40 -05:00
Kyle McMartin
f671c45df2 [PARISC] Arch-specific compat signals
Add enough arch-specific compat signals code to enable parisc64
to compile and boot out of the mainline tree. There are likely still
many dragons here, but this is a start to squashing the last
big difference between the mainline tree and the parisc-linux tree.
The remaining bugs can be squashed as they come up.

Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-22 20:57:42 -05:00
Helge Deller
81a3de3efd [PARISC] Use DEBUG_KERNEL to catch used-after-free __init data
Use CONFIG_DEBUG_KERNEL to catch kernel code which tries to access
__init data after it is freed. When CONFIG_DEBUG_KERNEL is not set
this also cleans up a WARN_ON at boot time. Also remove some dead
code from mm/init.c

Signed-off-by: Helge Deller <deller@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-22 20:26:55 -05:00
Helge Deller
1bcdd85482 [PARISC] Add CONFIG_DEBUG_RODATA to protect read-only data
Add the parisc version of the "mark rodata section read only" patches.
Based on code from and Signed-off-by Arjan van de Ven
<arjan@infradead.org>, Ingo Molnar <mingo@elte.hu>, Andi Kleen <ak@muc.de>,
Andrew Morton <akpm@osdl.org>, Linus Torvalds <torvalds@osdl.org>.

Signed-off-by: Helge Deller <deller@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-22 20:26:54 -05:00
Matthew Wilcox
a2bb214dcd [PARISC] Remove {,un}lock_kernel from perf ioctl
Remove the lock_kernel/unlock_kernel pair in the ioctl method. It
plainly wasn't protecting anything.

Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-22 20:26:52 -05:00
Kyle McMartin
85509c0007 [PARISC] Add chassis_power_off routine
Define a chassis_power_off routine that machines which have a way
to turn off the power supply can hook into. Formerly they were
using pm_power_off, which is now being used by generic code. Make
lasi.c use chassis_power_off instead of pm_power_off.

Note, all machines need to call machine_power_off so that the
switch can power off the machine, though halt -p may not necessarily
be able to work properly on the machine.

Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-22 20:26:50 -05:00
Kyle McMartin
526110f8c8 [PARISC] Remove obsolete _hlt cruft
Remove a bunch of obsolete code from process.c, these routines were
likely imported from the i386 version of process.c when the port
started. The routines are only used in floppy.c, which I seriously
doubt will ever work on parisc, due to architectural assumptions.

Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-22 20:26:48 -05:00
Kyle McMartin
9073315bbc [PARISC] Move pm_power_off export to process.c
Move the EXPORT_SYMBOL() of pm_power_off from parisc_ksyms.c to the
location of its definition in process.c

Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-22 20:26:46 -05:00
Helge Deller
96629c0b11 [PARISC] Clean up compiler warning in pci.c
Avoid compiler warning for unused variables on 32bit kernels by
conditionalizing the local variables on CONFIG_64BIT. PCI_HOST_ADDR()
only uses the hba argument on 64bit compiles.

Signed-off-by: Helge Deller <deller@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-22 20:26:43 -05:00
Kyle McMartin
10992092a8 [PARISC] Use F_EXTEND() for COMMAND_GLOBAL
Instead of wrapping the define of COMMAND_GLOBAL in #ifdef __LP64__
use the F_EXTEND() macro defined in asm/io.h, which is the preferred
way of extending mmio space addresses for either 32 or 64 bit machines.

Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-22 20:26:41 -05:00
James Bottomley
370361f884 [PARISC] Fix floating point invalid exception trap handler
Fix our trap handler to issue the correct floating point exception
for both types of invalid trap.

Signed-off-by: James Bottomley <jejb@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-22 20:26:39 -05:00
Helge Deller
c475eea292 [PARISC] Drop unused do_check_pgt_cache()
Drop the unused do_check_pgt_cache routine from mm/init.c and its
prototype in asm/pgalloc.h

Signed-off-by: Helge Deller <deller@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-22 20:26:37 -05:00
Grant Grundler
2c9aadabf4 [PARISC] Remove unnecessary extern declarations from asm/pci.h
Remove two unnecessary extern declarations from asm/pci.h.
They collide with what gcc4.0 assumed was static (and should be static).
Found by Joel Soete.

Signed-off-by: Grant Grundler <grundler@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-22 20:26:34 -05:00
Helge Deller
cb6fc18e9c [PARISC] Use kzalloc and other janitor-style cleanups
Helge,
  o Convert a bunch of kmalloc/memset uses to kzalloc.
  o pci.c: Add some __read_mostly annotations.
  o pci.c: Move constant pci_post_reset_delay to asm/pci.h
  o grfioctl.h: Add A4450A to comment of CRT_ID_VISUALIZE_EG.
  o Add some consts to perf.c/perf_images.h

Matthew,
  o sticore.c: Add some consts to suppress compile warnings.

Signed-off-by: Helge Deller <deller@parisc-linux.org>
Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-22 20:26:31 -05:00
Matthew Wilcox
a1c7444395 [PARISC] Update b180_defconfig
Update b180_defconfig to be more usable on other similar machines.
Enabling Lasi 82596, Harmony, Mux console, CCIO, HPPB, etc., means
this config is suitable for not only BXXX machines, but also CXXX
and JXXX class machines.

Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-22 20:26:29 -05:00
Matthew Wilcox
ce33941f02 [PARISC] Make flush_tlb_all_local take a void *
Make flush_tlb_all_local take a void * so it doesn't have to be cast
when using on_each_cpu().  This becomes a problem when on_each_cpu
is a macro.

Also remove the prototype of flush_tlb_all_local from .c files.

Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-21 20:46:31 -05:00
Jesse Brandeburg
35ec56bb78 [PATCH] e1000: Added disable packet split capability
Adds the ability to disability packet split at compile time and use the legacy receive path on PCI express hardware.  Made this a CONFIG option and modified the Kconfig, to reflect the new option.

Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: John Ronciak <john.ronciak@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
2006-01-18 16:17:57 -05:00
Russell King
ad3ed31c68 [PATCH] Add parisc_bus_type probe and remove methods
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Acked-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2006-01-13 11:26:06 -08:00
Al Viro
40f1f0deaf [PATCH] parisc: task_stack_page(), task_thread_info()
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-12 09:08:55 -08:00
Randy Dunlap
a941564458 [PATCH] capable/capability.h (arch/)
arch: Use <linux/capability.h> where capable() is used.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-11 18:42:14 -08:00
Matthew Wilcox
58b6c58cae [PARISC] Use STABS_DEBUG macro from vmlinux.lds.h
Cleanup vmlinux.lds.S by using STABS_DEBUG macro from vmlinux.lds.h
instead of repeating the sections.

Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-10 21:53:19 -05:00
Matthew Wilcox
f45adcf977 [PARISC] Fix Dino reporting on J2240
Fix Dino reporting on J2240. This particular machine thought it
had a Cujo. Also add J2240 Dino chip to the hp_hardware_list.

Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-10 21:53:14 -05:00
Grant Grundler
5cdb8205e9 [PARISC] Fix BLK_BOUNCE_HIGH on parisc by initializing max_low_pfn
max_low_pfn was not being set in arch/parisc/mm/init.c, causing severe
problems whenever anything tried to use BLK_BOUNCE_HIGH. Set it to
max_pfn like other similar architectures do.

Signed-off-by: Grant Grundler <grundler@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-10 21:52:11 -05:00
Matthew Wilcox
a01c8cb126 [PARISC] Fix GSC graphics cards with 64MB regions
Make knapps work with its 64MB gfx card.  I probably just broke another
machine in the process, but assuming 64MB when 64MB aligned is probably
safer than assuming 32MB all the time.

Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-10 21:51:33 -05:00
Kyle McMartin
e0565a1c83 [PARISC] Fix and cleanup ioremap.c to work with 4level-fixup.h
Fixup ioremap a bit. It seems to work on 32-bit kernels, but fails
miserably on the first ioremapped access on 64-bit kernels. Also, having
STI enabled causes it to fail. Probably because we're passing an ioremapped
region to a real-mode STI call...

Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-10 21:51:25 -05:00
Matthew Wilcox
1b2425e3c7 [PARISC] Make local cache flushes take a void *
Make flush_data_cache_local, flush_instruction_cache_local and
flush_tlb_all_local take a void * so they don't have to be cast
when using on_each_cpu().  This becomes a problem when on_each_cpu
is a macro (as it is in current -mm).

Also move the prototype of flush_tlb_all_local into tlbflush.h and
remove its declaration from .c files.

Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-10 21:49:21 -05:00
Helge Deller
8039de10aa [PARISC] Add __read_mostly section for parisc
Flag a whole bunch of things as __read_mostly on parisc. Also flag a few
branches as unlikely() and cleanup a bit of code.

Signed-off-by: Helge Deller <deller@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2006-01-10 20:35:03 -05:00
Christoph Hellwig
e6a6d2efcb [PATCH] sanitize building of fs/compat_ioctl.c
Now that all these entries in the arch ioctl32.c files are gone [1], we can
build fs/compat_ioctl.c as a normal object and kill tons of cruft.  We need a
special do_ioctl32_pointer handler for s390 so the compat_ptr call is done.
This is not needed but harmless on all other architectures.  Also remove some
superflous includes in fs/compat_ioctl.c

Tested on ppc64.

[1] parisc still had it's PPP handler left, which is not fully correct
    for ppp and besides that ppp uses the generic SIOCPRIV ioctl so it'd
    kick in for all netdevice users.  We can introduce a proper handler
    in one of the next patch series by adding a compat_ioctl method to
    struct net_device but for now let's just kill it - parisc doesn't
    compile in mainline anyway and I don't want this to block this
    patchset.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Matthew Wilcox <willy@debian.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-10 08:01:33 -08:00
Christoph Hellwig
ec3cad9690 [PATCH] move rtc compat ioctl handling to fs/compat_ioctl.c
This patch implements generic handling of RTC_IRQP_READ32, RTC_IRQP_SET32,
RTC_EPOCH_READ32 and RTC_EPOCH_SET32 in fs/compat_ioctl.c.  It's based on the
x86_64 code which needed a little massaging to be endian-clean.

parisc used COMPAT_IOCTL or generic w_long handlers for these whichce is wrong
and can't work because the ioctls encode sizeof(unsigned long) in their ioctl
number.  parisc also duplicated COMPAT_IOCTL entries for other rtc ioctls
which I remove in this patch, too.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Matthew Wilcox <matthew@wil.cx>
Acked-by: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-10 08:01:32 -08:00
Matt Mackall
e585e47031 [PATCH] tiny: Make *[ug]id16 support optional
Configurable 16-bit UID and friends support

This allows turning off the legacy 16 bit UID interfaces on embedded platforms.

   text    data     bss     dec     hex filename
3330172  529036  190556 4049764  3dcb64 vmlinux-baseline
3328268  529040  190556 4047864  3dc3f8 vmlinux

From: Adrian Bunk <bunk@stusta.de>

    UID16 was accidentially disabled for !EMBEDDED.

Signed-off-by: Matt Mackall <mpm@selenic.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-08 20:14:11 -08:00
Grant Grundler
9d7d57567c [PARISC] Remove unused variable in signal.c
Remove unused variable "struct siginfo si" in signal.c

Signed-off-by: Grant Grundler <grundler@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-11-17 16:43:52 -05:00
Christoph Hellwig
784412f74c [PARISC] remove drm compat ioctls handlers
Remove drm compat_ioctl handlers. The drm drivers have proper
compat_ioctl methods these days.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-11-17 16:41:26 -05:00
Christoph Hellwig
ad7dd338fb [PARISC] move PA perf driver over to ->compat_ioctl
Move PA perf driver over to ->compat_ioctl.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Randolph Chung <tausq@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-11-17 16:40:31 -05:00
Matthew Wilcox
83aceb5b6a [PARISC] Fix some compile problems in ptrace.c
Fix some compile problems:
- ret wasn't being initialised in all code paths
- I'm pretty sure 'goto out' should have been 'goto out_tsk'

Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-11-17 16:37:24 -05:00
Matthew Wilcox
4269b0d371 [PARISC] Improve the error message when we get a clashing mod path
Improve the error message when we get a clashing mod path, and
actually display the IODC data and path for the conflicting device.

Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-11-17 16:33:56 -05:00
Matthew Wilcox
ba5c4f1bae [PARISC] Return PDC_OK when alloc_pa_dev fails to enumerate all devices
Return PDC_OK when device registration fails so that we enumerate all
subsequent devices, even when we get two devices with the same hardware
path (which should never happen, but does with at least one revision of
rp8400 firmware).

Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-11-17 16:33:29 -05:00
Carlos O'Donell
aa0eecb07f [PARISC] Document some register usages in assembly files
Document clobbers and args in entry.S and syscall.S.

entry.S: Add comment to indicate that cr27 may recycle and EDEADLOCK
detection is not 100% correct. Since this is only enabled when using
ENABLE_LWS_DEBUG, the user is warned by the comment.

Signed-off-by: Carlos O'Donell <carlos@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-11-17 16:32:46 -05:00
Ryan Bradetich
75be99a8c5 [PARISC] Make redirecting irq messages less noisy
Make the "redirecting irq" message to not display on the console by
setting the severity to KERN_DEBUG.  The console was basically unusable.

Signed-off-by: Ryan Bradetich <rbrad@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-11-17 16:29:50 -05:00
Grant Grundler
03afe22f07 [PARISC] irq_affinityp[] only available for SMP builds
irq_affinityp[] only available for SMP builds, make code that uses
it conditional on CONFIG_SMP.

Signed-off-by: Grant Grundler <grundler@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-11-17 16:29:16 -05:00
James Bottomley
c2ab64d098 [PARISC] Add IRQ affinities
This really only adds them for the machines I can check SMP on, which
is CPU interrupts and IOSAPIC (so not any of the GSC based machines).

With this patch, irqbalanced can be used to maintain irq balancing.
Unfortunately, irqbalanced is a bit x86 centric, so it doesn't do an
incredibly good job, but it does work.

Signed-off-by: James Bottomley <jejb@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-11-17 16:28:37 -05:00
Kyle McMartin
1d4c452a85 [PARISC] Fix uniprocessor build by dummying smp_send_all_nop()
Since irq.c uses smp_send_all_nop, we must define it for UP builds
as well. Make it a static inline so it gets optimized away. This forces
irq.c to include <asm/smp.h> though.

Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-11-17 16:27:44 -05:00
James Bottomley
d911aed8ad [PARISC] Fix our interrupts not to use smp_call_function
Fix our interrupts not to use smp_call_function

On K and D class smp, the generic code calls this under an irq
spinlock, which causes the WARN_ON() message in smp_call_function()
(and is also illegal because it could deadlock).

The fix is to use a new scheme based on the IPI_NOP.

Signed-off-by: James Bottomley <jejb@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-11-17 16:27:02 -05:00
Grant Grundler
3f902886a8 [PARISC] Disable nesting of interrupts
Disable nesting of interrupts - still has holes

The offending sequence starts out like this:
1) take external interrupt
2) set_eiem() to only allow TIMER_IRQ; local interrupts still disabled
3) read the EIRR to get a "list" of pending interrupts
4) clear EIRR of pending interrupts we intend to handle
5) call __do_IRQ() to handle IRQ.
6) handle_IRQ_event() enables local interrupts (I-Bit)
7) take a timer interrupt
8) read EIRR to get a new list of pending interrupts
9) clear EIRR of pending interrupts we just read
10) handle pending interrupts found in (8)
11) set_eiem(cpu_eiem) and return
        [ TROUBLE! all enabled CPU IRQs are unmasked. }
12) handle remaining interrupts pending from (3)
        e.g. call __do_IRQ() -> handle_IRQ_event()..etc
        [ TROUBLE! call to handle_IRQ_event() can now enable *any* IRQ. }
13) set_eiem(cpu_eiem) and return

The problem is we now get into ugly race conditions with Timer and IPI
interrupts at this point.  I'm not exactly sure what happens when
things go wrong (perhaps nest calls to IPI or timer interrupt?).
But I'm certain it's not good.

This sequence will break sooner if (10) would accidentally leave
interrupts enabled.

I'm pretty sure the right answer is now to make cpu_eiem
a per CPU variable since all external interrupts on parisc
are per CPU. This means we will NOT need to send an IPI to
every CPU in the system when enabling or disabling an IRQ
since only one CPU needs to change it's EIEM.

Thanks to James Bottomley for (once again) pointing out the problem.

Signed-off-by: Grant Grundler <grundler@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-11-17 16:26:20 -05:00
James Bottomley
9a8b458406 [PARISC] Make sure timer and IPI execute with interrupts disabled
Fix a longstanding smp bug

The problem is that both the timer and ipi interrupts are being called
with interrupts enabled, which isn't what anyone is expecting.

The IPI issue has just started to show up by causing a BUG_ON in the
slab debugging code.  The timer issue never shows up because there's an
eiem work around in our irq.c

The fix is to label both these as SA_INTERRUPT which causes the generic
irq code not to enable interrupts.

I also suspect the smp_call_function timeouts we're seeing might be
connected with the fact that we disable IPIs when handling any other
type of interrupt.  I've put a WARN_ON in the code for executing
smp_call_function() with IPIs disabled.

Signed-off-by: James Bottomley <jejb@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2005-11-17 16:24:52 -05:00