Commit Graph
640 Commits
Author SHA1 Message Date
Ralf Baechle ece8a9e4f0 [MIPS] SMP: Fix use of cpumasks.
Noticed by Nick Piggin <[email protected]>.

Signed-off-by: Ralf Baechle <[email protected]>
2007-10-13 00:53:01 +01:00
Ralf Baechle bdf5d42c6e [MIPS] VPE: reimplement ELF loader.
Loading ELF binaries based on the section table is totally wrong.  This
still leaves the other fat bug of referencing symbols in an executable
unfixed, so people better don't run strip on their binaries ...

As added bonus the new loader is also 23 lines shorter.

Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:19 +01:00
Franck Bui-Huu dec8b1ca99 [MIPS] Add BUG_ON assertion for attempt to run kernel on the wrong CPU type.
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:18 +01:00
Ralf Baechle 89a8a5a6c9 [MIPS] SMP: Use ISO C struct initializer for local structs.
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:18 +01:00
Ralf Baechle c50cade95b [MIPS] SMP: Kill useless casts.
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:18 +01:00
Ralf Baechle b5eb551145 [MIPS] Kill num_online_cpus() loops.
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:18 +01:00
Ralf Baechle bd6aeeffcc [MIPS] SMP: Implement smp_call_function_mask().
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:18 +01:00
Ralf Baechle 9966db25de [MIPS] Make facility to convert CPU types to strings generally available.
So far /proc/cpuinfo has been the only user but human readable processor
name are more useful than that for proc.

Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:17 +01:00
Ralf Baechle 21a151d8ca [MIPS] checkfiles: Fix "need space after that ','" errors.
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:15 +01:00
Ralf Baechle 49a89efbbb [MIPS] Fix "no space between function name and open parenthesis" warnings.
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:15 +01:00
Ralf Baechle 10cc352907 [MIPS] Allow hardwiring of the CPU type to a single type for optimization.
This saves a few k on systems which only ever ship with a single CPU type.

Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:15 +01:00
Ralf Baechle d865bea4da [MIPS] i8253 PIT clocksource and clockevent drivers
Derived from the i386 variant with a few x86 complexities chopped off.

Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:10 +01:00
Ralf Baechle ea5804015c [MIPS] Dyntick support for SMTC:
The kernel currently only supports broadcasting of the timer interrupt
from a single timer, not multicasting into two multicast groups of
processors.  So the implemented mechanism for SMTC works by broadcasting
the cp0 compare interrupt on VPE 0 and ignoring it on any additional VPEs.

Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:09 +01:00
Ralf Baechle 7bcf7717b6 [MIPS] Implement clockevents for R4000-style cp0 count/compare interrupt
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:09 +01:00
Ralf Baechle 91a2fcc886 [MIPS] Consolidate all variants of MIPS cp0 timer interrupt handlers.
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:09 +01:00
Ralf Baechle 90b02340dc [MIPS] Switch from to_tm to rtc_time_to_tm
This replaces the MIPS-specific to_tm function with the generic
rtc_time_to_tm function.  The big difference between the two functions is
that rtc_time_to_tm uses epoch 70 while to_tm uses 1970, so the result of
rtc_time_to_tm needs to be fixed up.

Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:09 +01:00
Ralf Baechle 4b550488f8 [MIPS] Deforest the function pointer jungle in the time code.
Hard to follow who is pointing what to where and why so it's simply getting
in the way of the time code renovation.

Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:08 +01:00
Ralf Baechle f5ff0a2802 [MIPS] Use generic NTP code for all MIPS platforms
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:08 +01:00
Ralf Baechle 05dc8c02bf [MIPS] ARC: Get rid of mips_machgroup
This has not been any serious user of this ill conceived thing since the
original invention in like '95.

Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:08 +01:00
Ralf Baechle 641e97f318 [MIPS] Sibyte: Replace SB1 cachecode with standard R4000 class cache code.
It may not be perfect yet but the SB1 code is badly borken and has
horrible performance issues.

Downside: This seriously breaks support for pass 1 parts of the BCM1250
where indexed cacheops don't work quite reliable but I seem to be the
last one on the planet with a pass 1 part anyway.

Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:05 +01:00
Maciej W. Rozycki e5d77754c5 [MIPS] R3000 setup for kernel_thread()
Match the R4000 semantics for the initial state of interrupt/kernel
status register flags for the R3000 in kernel_thread().

Signed-off-by: Maciej W. Rozycki <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:04 +01:00
Sam Ravnborg 0f5c906442 [MIPS] Introduce a consistent style for vmlinux.lds.
This style will be consitent with all other arch's - soon.

In addition:
- Moved a few labels inside brackets for the sections they specify
  to prevent that linker alignmnet made them point before the section start

Signed-off-by: Sam Ravnborg <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:04 +01:00
Yoichi Yuasa d5ab1a6910 [MIPS] Add GT641xx IRQ routines.
Signed-off-by: Yoichi Yuasa <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:04 +01:00
Yoichi Yuasa d80c1c0b22 [MIPS] i8295 cleanups.
o Move i8259 function declarations to include/asm-mips/i8259.h
 o Make i8259.c functions static where possible.

Signed-off-by: Yoichi Yuasa <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:04 +01:00
Ralf Baechle ec70f65e3b [MIPS] Kill useless volatile keyword
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:03 +01:00
Thomas Gleixner 4e45171c4e [MIPS] cleanup struct irqaction initializers
Signed-off-by: Thomas Gleixner <[email protected]>
CC: Ralf Baechle <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:02 +01:00
Aurelien Jarno 1c0c13eb93 [MIPS] Add support for BCM47XX CPUs.
Note that the BCM4710 does not support the wait instruction, this
is not a mistake in the code.
    
It originally comes from the OpenWrt patches.
    
Cc: Michael Buesch <[email protected]>
Cc: Felix Fietkau <[email protected]>
Cc: Florian Schirmer <[email protected]>
Signed-off-by: Aurelien Jarno <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:46:02 +01:00
Ralf Baechle d87d0c930a [MIPS] SMTC: Microoptimize atomic_postincrement for non-weak consistency.
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:45:58 +01:00
Kevin D. Kissell f571eff0a2 [MIPS] IRQ Affinity Support for SMTC on Malta Platform
Signed-off-by: Kevin D. Kissell <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-11 23:45:57 +01:00
Maciej W. Rozycki 6f6b3940ed [MIPS] vmlinux.lds.S: Handle note sections
Store any note sections after the exception tables like the other
architectures do.  This is required for .note.gnu.build-id emitted from
binutils 2.18 onwards if nothing else.

Signed-off-by: Maciej W. Rozycki <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-10-01 14:17:50 +01:00
Kyle McMartin 422efb17ee [MIPS] Fix fallocate on o32 binary compat ABI
MIPS was mistakenly forgetting to use the fallocate compat wrapper, which
I noticed while cleaning up all the duplicate fallocate wrappers.

Signed-off-by: Kyle McMartin <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-09-28 19:27:20 +01:00
Ralf Baechle 1146fe3050 [MIPS] SMTC: Make ack_bad_irq() safe with no IM backstop.
Issue reported and original patch by Kevin Kissel, cleaner (imho)
implementation by me.

Signed-off-by: Ralf Baechle <[email protected]>
2007-09-24 18:13:02 +01:00
Maciej W. Rozycki 09abbcffb3 [MIPS] cpu-bugs64.c: GCC 3.3 constraint workaround
Add a workaround to address warnings generated on the "n" constraint by
GCC 3.3 and below.

Signed-off-by: Maciej W. Rozycki <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-09-19 19:33:14 +01:00
Ralf Baechle 50da469a79 [MIPS] 20Kc: Disable use of WAIT instruction.
Another issue with 20Kc's WAIT, waiting for more details.  With the
2.6.23 release immindent simply disable the use of WAIT instead of a
more fancy workaround.

Signed-off-by: Ralf Baechle <[email protected]>
2007-09-14 19:08:43 +01:00
Yoichi Yuasa b1043cc7d9 [MIPS] Add #include <linux/profile.h> to arch/mips/kernel/time.c
It refer to CPU_PROFILING.

arch/mips/kernel/time.c: In function 'local_timer_interrupt':
arch/mips/kernel/time.c:142: error: implicit declaration of function 'profile_tick'
arch/mips/kernel/time.c:142: error: 'CPU_PROFILING' undeclared (first use in this function)
arch/mips/kernel/time.c:142: error: (Each undeclared identifier is reported only once
arch/mips/kernel/time.c:142: error: for each function it appears in.)

Signed-off-by: Yoichi Yuasa <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-09-14 19:08:42 +01:00
Johannes Dickgreber 82411390a2 [MIPS] N32 needs to use compat_sys_futimesat
Signed-off-by: Ralf Baechle <[email protected]>
2007-09-14 19:08:42 +01:00
Ralf Baechle 4d24c8f955 [MIPS] rtlx: Fix build error.
CC      arch/mips/kernel/rtlx.o
cc1: warnings being treated as errors
arch/mips/kernel/rtlx.c:59: warning: 'irq' defined but not used
arch/mips/kernel/rtlx.c:60: warning: 'irq_num' defined but not used

Signed-off-by: Ralf Baechle <[email protected]>
2007-09-14 19:08:42 +01:00
Ralf Baechle e606c109c4 [MIPS] rtlx: fix int vs. long bug.
CC      arch/mips/kernel/rtlx.o
arch/mips/kernel/rtlx.c: In function 'rtlx_init':
arch/mips/kernel/rtlx.c:114: warning: format '%x' expects type 'unsigned int', but argument 3 has type 'long unsigned int'

Signed-off-by: Ralf Baechle <[email protected]>
2007-09-14 19:08:42 +01:00
Kyungmin Park d77a283d9b [MIPS] i8259: Add disable method.
After 76d2160147, the qemu NE2000 was
frequently producing WATCHDOG timeouts.

Signed-off-by: Kyungmin Park <[email protected]>
Cc: Ralf Baechle <[email protected]>
Cc: Ingo Molnar <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-09-10 21:25:27 +01:00
Thiemo Seufer 948a34cf39 [MIPS] Maintain si_code field properly for FP exceptions
The appended patch adds code to update siginfo_t's si_code field. It
fixes e.g. a floating point overflow regression in the SBCL testsuite.

Signed-off-By: Thiemo Seufer <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-08-27 02:16:59 +01:00
Thiemo Seufer 34412c7231 [MIPS] SMTC: Fix duplicate status dumps on NMI
Also removes the while(1); loop by propagating the ATTRIB_NORET of die()
to nmi_exception_handler.

Signed-off-by: Thiemo Seufer <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-08-27 02:16:59 +01:00
Nicolas Schichan 97ce9a8d6c [MIPS] Unconditionally writeback and invalidate caches on kexec.
Signed-off-by: Nicolas Schichan <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-08-27 02:16:59 +01:00
Ralf Baechle c99dcac435 [MIPS] Fix gcc 3.3 warning.
CC      arch/mips/kernel/cpu-bugs64.o
arch/mips/kernel/cpu-bugs64.c: In function 'align_mod':
arch/mips/kernel/cpu-bugs64.c:23: warning: asm operand 0 probably doesn't match constraints
arch/mips/kernel/cpu-bugs64.c:23: warning: asm operand 1 probably doesn't match constraints

Signed-off-by: Ralf Baechle <[email protected]>
2007-08-27 02:16:58 +01:00
Ilpo Järvinen 0806133af4 [MIPS] Fix invalid semicolon after if statement
Signed-off-by: Ilpo Järvinen <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-08-27 02:16:58 +01:00
Ralf Baechle bdd3ede0e1 [MIPS] MT: Use kallsyms in CPU state dump
Signed-off-by: Ralf Baechle <[email protected]>
2007-08-27 02:16:56 +01:00
Ralf Baechle fe56b954ea [MIPS] SMTC: Move MIPS_CPU_IPI_IRQ definition into header.
Signed-off-by: Ralf Baechle <[email protected]>
2007-08-27 02:16:55 +01:00
Ralf Baechle 428ab280a0 [MIPS] SMP: Scatter __cpuinit over the code as needed.
MIPS doesn't do CPU hotplugging yet but since many of the functions don't
even have an __init let's fix this right.

Signed-off-by: Ralf Baechle <[email protected]>
2007-08-27 02:16:54 +01:00
Atsushi Nemoto 5d3fdeacee [MIPS] Remove dead code from irq_txx9.c
Signed-off-by: Atsushi Nemoto <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-08-27 02:16:54 +01:00
Daniel Jacobowitz 6c76988b81 [MIPS] Do not discard extra debugging sections.
Leaving these sections is useful to some tools that look at the image, and
none of them are loaded into memory.  The .mdebug.abi64 section, in
particular, lets GDB recognize vmlinux.32 as an N64 program instead of
guessing that it is O32.

Signed-off-by: Daniel Jacobowitz <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-08-27 02:16:53 +01:00
Atsushi Nemoto 8420fd00e8 [MIPS] The irq_chip for TX39/TX49 SoCs
Add generic irq_chip for TX39/TX49 SoCs.  This can be replace
jmr3927_irq_irc, tx4927_irq_pic_type and tx4938_irq_pic_type.

Signed-off-by: Atsushi Nemoto <[email protected]>
Signed-off-by: Ralf Baechle <[email protected]>
2007-08-27 02:16:52 +01:00