The sparc32 version of arch_write_unlock() is just a plain assignment.
Unfortunately this allows the compiler to schedule side-effects in a
protected region to occur after the HW-level unlock, which is broken.
E.g., the following trivial test case gets miscompiled:
#include <linux/spinlock.h>
rwlock_t lock;
int counter;
void foo(void) { write_lock(&lock); ++counter; write_unlock(&lock); }
Fixed by adding a compiler memory barrier to arch_write_unlock(). The
sparc64 version combines the barrier and assignment into a single asm(),
and implements the operation as a static inline, so that's what I did too.
Compile-tested with sparc32_defconfig + CONFIG_SMP=y.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
The sparc64 spinlock_64.h contains a number of operations defined
first as static inline functions, and then as macros with the same
names and parameters as the functions. Maybe this was needed at
some point in the past, but now nothing seems to depend on these
macros (checked with a recursive grep looking for ifdefs on these
names). Other archs don't define these identity-macros.
So this patch deletes these unnecessary macros.
Compile-tested with sparc64_defconfig.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
Building kernel 3.0 for an n2100 (plat-iop) results in:
In file included from arch/arm/plat-iop/cp6.c:20:
/tmp/linux-3.0/arch/arm/include/asm/traps.h:12: warning: 'struct pt_regs' declared inside parameter list
/tmp/linux-3.0/arch/arm/include/asm/traps.h:12: warning: its scope is only this definition or declaration, which is probably not what you want
/tmp/linux-3.0/arch/arm/include/asm/traps.h:48: warning: 'struct pt_regs' declared inside parameter list
/tmp/linux-3.0/arch/arm/include/asm/traps.h:48: warning: 'struct task_struct' declared inside parameter list
arch/arm/plat-iop/cp6.c:45: warning: initialization from incompatible pointer type
Nothing here depends on the layout of pt_regs or task_struct, so this
can be fixed by adding forward struct declarations to asm/traps.h.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Russell King <[email protected]>
This cleans up a few bits in binfmt_elf.c and binfmts.h:
- the hasvdso field in struct linux_binfmt is unused, so remove it and
the only initialization of it
- the elf_map CPP symbol is not defined anywhere in the kernel, so
remove an unnecessary #ifndef elf_map
- reduce excessive indentation in elf_format's initializer
- add missing spaces, remove extraneous spaces
No functional changes, but tested on x86 (32 and 64 bit), powerpc (32 and
64 bit), sparc64, arm, and alpha.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
The 2.6.36-rc kernel added three new system calls:
fanotify_init, fanotify_mark, and prlimit64. This
patch wires them up on Alpha.
Built and booted on an XP900. Untested beyond that.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Matt Turner <[email protected]>
The 2.6.36-rc kernel added three new system calls:
fanotify_init, fanotify_mark, and prlimit64. This patch
wires them up on ARM.
The only non-trivial issue here is the u64 argument to
sys_fanotify_mark(), but it is the 3rd argument and thus
passed in r2/r3 in both kernel and user space, so it causes
no problems.
Tested with a 2.6.36-rc2 EABI kernel on an ixp4xx machine.
Tested-by: Anand Gadiyar <[email protected]>
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Russell King <[email protected]>
sys_accept4() was added in kernel 2.6.28, but ARM was not updated
to include it. The number and types of parameters is such that
no ARM-specific processing is needed, so wiring up sys_accept4()
just requires defining __NR_accept4 and adding a direct call in
the syscall entry table.
Tested with an EABI 2.6.35 kernel and Ulrich Drepper's original
accept4() test program, modified to define __NR_accept4 for ARM.
Using the updated unistd.h also eliminates a warning then building
glibc (2.10.2 and newer) about accept4() being unimplemented.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Russell King <[email protected]>
The kernel's math-emu code contains a macro _FP_FROM_INT() which is
used to convert an integer to a raw normalized floating-point value.
It does this basically in three steps:
1. Compute the exponent from the number of leading zero bits.
2. Downshift large fractions to put the MSB in the right position
for normalized fractions.
3. Upshift small fractions to put the MSB in the right position.
There is an boundary error in step 2, causing a fraction with its
MSB exactly one bit above the normalized MSB position to not be
downshifted. This results in a non-normalized raw float, which when
packed becomes a massively inaccurate representation for that input.
The impact of this depends on a number of arch-specific factors,
but it is known to have broken emulation of FXTOD instructions
on UltraSPARC III, which was originally reported as GCC bug 44631
<http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44631>.
Any arch which uses math-emu to emulate conversions from integers to
same-size floats may be affected.
The fix is simple: the exponent comparison used to determine if the
fraction should be downshifted must be "<=" not "<".
I'm sending a kernel module to test this as a reply to this message.
There are also SPARC user-space test cases in the GCC bug entry.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
A __naked function is defined in C but with a body completely implemented
by asm(), including any prologue and epilogue. These asm() bodies expect
standard calling conventions for parameter passing. Older GCCs implement
that correctly, but 4.[56] currently do not, see GCC PR44290. In the
Linux kernel this breaks ARM, causing most arch/arm/mm/copypage-*.c
modules to get miscompiled, resulting in kernel crashes during bootup.
Part of the kernel fix is to augment the __naked function attribute to
also imply noinline and noclone. This patch implements that, and has been
verified to fix boot failures with gcc-4.5 compiled 2.6.34 and 2.6.35-rc1
kernels. The patch is a no-op with older GCCs.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Khem Raj <[email protected]>
Cc: Russell King <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
2.6.34-rc1 added kernel/elfcore.c which includes <asm/elf.h>.
On ARM, this results in:
In file included from include/linux/elf.h:7,
from kernel/elfcore.c:1:
/tmp/linux-2.6.34-rc1/arch/arm/include/asm/elf.h:101: warning: 'struct task_struct' declared inside parameter list
/tmp/linux-2.6.34-rc1/arch/arm/include/asm/elf.h:101: warning: its scope is only this definition or declaration, which is probably not what you want
Including <linux/sched.h> seems a bit heavyweight, so this patch just
adds a tentative declaration of struct task_struct in <asm/elf.h>.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Russell King <[email protected]>
This fixes a futex key reference count bug in futex_lock_pi(),
where a key's reference count is incremented twice but decremented
only once, causing the backing object to not be released.
If the futex is created in a temporary file in an ext3 file system,
this bug causes the file's inode to become an "undead" orphan,
which causes an oops from a BUG_ON() in ext3_put_super() when the
file system is unmounted. glibc's test suite is known to trigger this,
see <http://bugzilla.kernel.org/show_bug.cgi?id=14256>.
The bug is a regression from 2.6.28-git3, namely Peter Zijlstra's
38d47c1b70 "[PATCH] futex: rely on
get_user_pages() for shared futexes". That commit made get_futex_key()
also increment the reference count of the futex key, and updated its
callers to decrement the key's reference count before returning.
Unfortunately the normal exit path in futex_lock_pi() wasn't corrected:
the reference count is incremented by get_futex_key() and queue_lock(),
but the normal exit path only decrements once, via unqueue_me_pi().
The fix is to put_futex_key() after unqueue_me_pi(), since 2.6.31
this is easily done by 'goto out_put_key' rather than 'goto out'.
Signed-off-by: Mikael Pettersson <[email protected]>
Acked-by: Peter Zijlstra <[email protected]>
Acked-by: Darren Hart <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Cc: <[email protected]>
Add a better sched_clock() to the ixp4xx platform,
implemented via its clocksource support.
This is based on the sched_clock() I implemented for
the IOP platform. Tested on a ds101 ixp420 machine.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Russell King <[email protected]>
When sata_promise encounters an overrun or underrun error it
translates that to a libata AC_ERR_HSM, causing a hard reset.
Since over/under-runs were thought to be rare and transient,
this action seemed reasonable.
Unfortunately it turns out that the controller throws overrun
errors when e.g. hal polls a CD or DVD writer containing blank
media, causing long sequences of hard resets and retries before
EH finally gives up.
This patch updates sata_promise to classify over/under-runs as
AC_ERR_OTHER instead. This allows libata EH and upper layers to
retry or fail the operation as they see fit without the disruption
caused by repeated hard resets.
This fixes a problem using a DVD-RAM drive with sata_promise,
reported by Thomas Schorpp. I also tested it on a DVD-RW drive.
Signed-off-by: Mikael Pettersson <[email protected]>
Tested-by: thomas schorpp <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
The x86 lapic nmi watchdog does not recognize AMD Family 11h,
resulting in:
NMI watchdog: CPU not supported
As far as I can see from available documentation (the BKDM),
family 11h looks identical to family 10h as far as the PMU
is concerned.
Extending the check to accept family 11h results in:
Testing NMI watchdog ... OK.
I've been running with this change on a Turion X2 Ultra ZM-82
laptop for a couple of weeks now without problems.
Signed-off-by: Mikael Pettersson <[email protected]>
Cc: Andreas Herrmann <[email protected]>
Cc: Joerg Roedel <[email protected]>
Cc: <[email protected]>
LKML-Reference: <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
This adds a better sched_clock() to the IOP platform,
implemented using its new clocksource support.
Tested on n2100, compile-tested for all plat-iop machines.
[[email protected]: allow early cp6 access]
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Dan Williams <[email protected]>
This updates the IOP platform to use the kernel's generic time
framework. With clockevent support in place, this reduces to
selecting GENERIC_TIME and removing the platform's private timer
->offset() operation (iop_gettimeoffset).
Tested on n2100, compile-tested for all plat-iop machines.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Dan Williams <[email protected]>
This updates the IOP platform to expose the interrupting
timer 0 as a clockevent object. The timer interrupt handler
is changed to call the clockevent ->event_handler() instead
of timer_tick(), and ->set_next_event() and ->set_mode()
operations are added to allow the mode of the timer to be
updated (required for ONESHOT/NOHZ mode).
Timer 0 must now be properly initialised, which requires
a new write_tcr0() function from the mach-specific code.
The mode of timer 0 must be read at the start of ->set_mode(),
which requires a new read_tmr0() function from the mach-
specific code.
Initial setup of timer 0 is also rewritten to be more robust.
Tested on n2100, compile-tested for all plat-iop machines.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Dan Williams <[email protected]>
This updates the IOP platform to expose the free-running
timer 1 as a clocksource object. This timer is now also
properly initialised, which requires a new write_tcr1()
function from the mach-specific code. Apart from the
explicit initialisation, there is no functional change
in how timer 1 is programmed.
Tested on n2100, compile-tested for all plat-iop machines.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Dan Williams <[email protected]>
Compiling the radeon DRM driver with !CONFIG_DEBUG_FS
throws the following warnings:
drivers/gpu/drm/radeon/radeon_ttm.c: In function 'radeon_ttm_debugfs_init':
drivers/gpu/drm/radeon/radeon_ttm.c:714: warning: unused variable 'i'
drivers/gpu/drm/radeon/radeon_ttm.c: At top level:
drivers/gpu/drm/radeon/radeon_ttm.c:692: warning: 'radeon_mem_types_list' defined but not used
drivers/gpu/drm/radeon/radeon_ttm.c:693: warning: 'radeon_mem_types_names' defined but not used
Fix: move these variables inside the #if defined(CONFIG_DEBUG_FS)
block in radeon_ttm_debugsfs_init(), which is the only place using them.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>
Compiling DRM throws the following warning if MAGIC_SYSRQ is disabled:
drivers/gpu/drm/drm_fb_helper.c:101: warning: 'sysrq_drm_fb_helper_restore_op' defined but not used
Fix: place sysrq_drm_fb_helper_restore_op and associated
definitions inside #ifdef CONFIG_MAGIC_SYSRQ.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>
The body of the mach-ixp4xx arch_idle() is mysteriously
disabled by an #if 0 .. #endif. Normally one would expect
to find a call to cpu_do_idle() there, but that call is
disabled, even though cpu_do_idle() is implemented for
XScale cores (and ixp4xx is one).
The explanation can be found in the ixp42x developer's manual
which states that the XScale core clock and power management
registers aren't implemented on ixp42x [3.5.2.2].
Also, the disabled code has suffered from bit rot:
- it checks hlt_counter which is obsolete, as that variable
and all related code now is private to kernel/process.c
- it passes too many parameters to cpu_do_idle()
So this patch:
- adds a comment before the #if 0 to explain why
cpu_do_idle() mustn't be called on ixp4xx
- removes the obsolete test of hlt_counter and the
obsolete parameter to cpu_do_idle()
This is purely a documentation fixup and changes no
generated code. Even so, it has been tested on an
ixp420 machine (ds101).
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Krzysztof Hałasa <[email protected]>
This patch does a few simple cleanups of the ixp4xx timer
and clocksource/clockevent code in mach-ixp4xx/common.c:
- ixp4xx_clocksource_init() is static and always returns 0,
which is ignored by its only caller: make it return void
- ixp4xx_clockevent_init(): ditto
- ixp4xx_get_cycles() is only referenced locally: make it static
- use the ixp4xx_timer_irq.dev_id field to pass &clockevent_ixp4xx
to ixp4xx_timer_interrupt() via its dev_id parameter, allowing
the code in ixp4xx_timer_interrupt() to be smaller and faster
Tested on an ixp420 machine (ds101).
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Krzysztof Hałasa <[email protected]>
The mv64xxx_i2c_intr() irq handler in drivers/i2c/busses/i2c-mv64xxx.c
is declared as returning 'int', resulting in this compile-time warning:
drivers/i2c/busses/i2c-mv64xxx.c: In function 'mv64xxx_i2c_probe':
drivers/i2c/busses/i2c-mv64xxx.c:540: warning: passing argument 2 of 'request_irq' from incompatible pointer type
Fix: correct the return type to 'irqreturn_t'.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Ben Dooks <[email protected]>
sata_promise's reset code has deviated quite a bit from
the Promise reference driver's, and it has been observed
to fail to recover from errors in some cases.
This patch thus updates the reset code to more closely
match the reference driver:
- soft reset (pdc_reset_port):
* wait for ATA engine to not be in packet command mode
(2nd gen only)
* write reset bit in PDC_CTLSTAT before the first read
in the loop
* for 2nd gen SATA follow up with FPDMA reset and clearing
error status registers
- hard reset (pdc_sata_hardreset):
* wait for ATA engine to not be in packet command mode
(2nd gen only)
* reset ATA engine via the PCI control register
* Tejun's change to use non-waiting hardreset + follow-up SRST
I'm not changing the hotplug mask bits since they are taken care
of by sata_promise's ->freeze() and ->thaw() operations. And I'm
not writing the PMP port # because that's always zero (for now).
Tested here on various controllers. In particular, one disk
which used to timeout and fail to recover from certain hdparm
and smartmonctl commands now works nicely.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
1st generation Promise SATA chips are prone to generating spurious
hotplug events which can disrupt normal operation. This has been
observed on 20376 and 20378 chips. This patch thus disables hotplug
support on 1st gen chips while leaving it enabled for 2nd gen chips.
The pdc_sata_hotplug_offset() function becomes redundant so it is
removed.
Tested on 1st gen 20376 and 20378 mainboard chips and on a 2nd gen
SATA300 PCI card.
Signed-off-by: Mikael Pettersson <[email protected]>
Tested-by: Kurt Roeckx <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
The txq_set_wrr() function in drivers/net/mv643xx_eth.c is
unused, not even referenced under #if 0 or something like that,
which results in a compile-time warning:
drivers/net/mv643xx_eth.c:1070: warning: 'txq_set_wrr' defined but not used
Fix: remove it.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
This patch adds support for TIF_RESTORE_SIGMASK to ARM's
signal handling, which allows to hook up the pselect6, ppoll,
and epoll_pwait syscalls on ARM.
Tested here with eabi userspace and a test program with a
deliberate race between a child's exit and the parent's
sigprocmask/select sequence. Using sys_pselect6() instead
of sigprocmask/select reliably prevents the race.
The other arch's support for TIF_RESTORE_SIGMASK has evolved
over time:
In 2.6.16:
- add TIF_RESTORE_SIGMASK which parallels TIF_SIGPENDING
- test both when checking for pending signal [changed later]
- reimplement sys_sigsuspend() to use current->saved_sigmask,
TIF_RESTORE_SIGMASK [changed later], and -ERESTARTNOHAND;
ditto for sys_rt_sigsuspend(), but drop private code and
use common code via __ARCH_WANT_SYS_RT_SIGSUSPEND;
- there are now no "extra" calls to do_signal() so its oldset
parameter is always ¤t->blocked so need not be passed,
also its return value is changed to void
- change handle_signal() to return 0/-errno
- change do_signal() to honor TIF_RESTORE_SIGMASK:
+ get oldset from current->saved_sigmask if TIF_RESTORE_SIGMASK
is set
+ if handle_signal() was successful then clear TIF_RESTORE_SIGMASK
+ if no signal was delivered and TIF_RESTORE_SIGMASK is set then
clear it and restore the sigmask
- hook up sys_pselect6() and sys_ppoll()
In 2.6.19:
- hook up sys_epoll_pwait()
In 2.6.26:
- allow archs to override how TIF_RESTORE_SIGMASK is implemented;
default set_restore_sigmask() sets both TIF_RESTORE_SIGMASK and
TIF_SIGPENDING; archs need now just test TIF_SIGPENDING again
when checking for pending signal work; some archs now implement
TIF_RESTORE_SIGMASK as a secondary/non-atomic thread flag bit
- call set_restore_sigmask() in sys_sigsuspend() instead of setting
TIF_RESTORE_SIGMASK
In 2.6.29-rc:
- kill sys_pselect7() which no arch wanted
So for 2.6.31-rc6/ARM this patch does the following:
- Add TIF_RESTORE_SIGMASK. Use the generic set_restore_sigmask()
which sets both TIF_SIGPENDING and TIF_RESTORE_SIGMASK, so
TIF_RESTORE_SIGMASK need not claim one of the scarce low thread
flags, and existing TIF_SIGPENDING and _TIF_WORK_MASK tests need
not be extended for TIF_RESTORE_SIGMASK.
- sys_sigsuspend() is reimplemented to use current->saved_sigmask
and set_restore_sigmask(), making it identical to most other archs
- The private code for sys_rt_sigsuspend() is removed, instead
generic code supplies it via __ARCH_WANT_SYS_RT_SIGSUSPEND.
- sys_sigsuspend() and sys_rt_sigsuspend() no longer need a pt_regs
parameter, so their assembly code wrappers are removed.
- handle_signal() is changed to return 0 on success or -errno.
- The oldset parameter to do_signal() is now redundant and removed,
and the return value is now also redundant and changed to void.
- do_signal() is changed to honor TIF_RESTORE_SIGMASK:
+ get oldset from current->saved_sigmask if TIF_RESTORE_SIGMASK
is set
+ if handle_signal() was successful then clear TIF_RESTORE_SIGMASK
+ if no signal was delivered and TIF_RESTORE_SIGMASK is set then
clear it and restore the sigmask
- Hook up sys_pselect6, sys_ppoll, and sys_epoll_pwait.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Russell King <[email protected]>
2.6.31-rc kernels don't boot on my ixp4xx box (ds101), because the libata
driver doesn't find the PCI IDE controller any more. 2.6.30 was fine.
I traced this to a PCI update (1f82de10d6)
in 2.6.30-git19. Diffing the kernel boot logs from 2.6.30-git18 and
2.6.30-git19 illustrates the breakage:
> --- dmesg-2.6.30-git18 2009-08-04 01:45:22.000000000 +0200
> +++ dmesg-2.6.30-git19 2009-08-04 01:45:46.000000000 +0200
> @@ -26,6 +26,13 @@
> pci 0000:00:02.2: PME# supported from D0 D1 D2 D3hot
> pci 0000:00:02.2: PME# disabled
> PCI: bus0: Fast back to back transfers disabled
> +pci 0000:00:01.0: BAR 0: can't allocate I/O resource [0x10000-0xffff]
> +pci 0000:00:01.0: BAR 1: can't allocate I/O resource [0x10000-0xffff]
> +pci 0000:00:01.0: BAR 2: can't allocate I/O resource [0x10000-0xffff]
> +pci 0000:00:01.0: BAR 3: can't allocate I/O resource [0x10000-0xffff]
> +pci 0000:00:01.0: BAR 4: can't allocate I/O resource [0x10000-0xffff]
> +pci 0000:00:02.0: BAR 4: can't allocate I/O resource [0x10000-0xffff]
> +pci 0000:00:02.1: BAR 4: can't allocate I/O resource [0x10000-0xffff]
> bio: create slab <bio-0> at 0
> SCSI subsystem initialized
> NET: Registered protocol family 2
> @@ -44,11 +51,7 @@
> console [ttyS0] enabled
> serial8250.0: ttyS1 at MMIO 0xc8001000 (irq = 13) is a XScale
> Driver 'sd' needs updating - please use bus_type methods
> -PCI: enabling device 0000:00:01.0 (0140 -> 0141)
> -scsi0 : pata_artop
> -scsi1 : pata_artop
> -ata1: PATA max UDMA/100 cmd 0x1050 ctl 0x1060 bmdma 0x1040 irq 28
> -ata2: PATA max UDMA/100 cmd 0x1058 ctl 0x1064 bmdma 0x1048 irq 28
> +pata_artop 0000:00:01.0: no available native port
> Using configured DiskOnChip probe address 0x50000000
> DiskOnChip found at 0x50000000
> NAND device: Manufacturer ID: 0x98, Chip ID: 0x73 (Toshiba NAND 16MiB 3,3V 8-bit)
The specific change in 1f82de10d6 responsible
for this failure turned out to be the following:
> --- a/drivers/pci/probe.c
> +++ b/drivers/pci/probe.c
> @@ -193,7 +193,7 @@ int __pci_read_base(struct pci_dev *dev, enum pci_bar_type type,
> res->flags |= pci_calc_resource_flags(l) | IORESOURCE_SIZEALIGN;
> if (type == pci_bar_io) {
> l &= PCI_BASE_ADDRESS_IO_MASK;
> - mask = PCI_BASE_ADDRESS_IO_MASK & 0xffff;
> + mask = PCI_BASE_ADDRESS_IO_MASK & IO_SPACE_LIMIT;
> } else {
> l &= PCI_BASE_ADDRESS_MEM_MASK;
> mask = (u32)PCI_BASE_ADDRESS_MEM_MASK;
Every arch except arm's ixp4xx defines IO_SPACE_LIMIT as an all-bits-one
bitmask, typically -1UL but sometimes only a 16-bit 0x0000ffff. But ixp4xx
defines it as 0xffff0000, which is now causing the PCI failures.
Russell King noted that ixp4xx has 64KB PCI IO space, so IO_SPACE_LIMIT
should be 0x0000ffff. This patch makes that change, which fixes the PCI
failures on my ixp4xx box.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Krzysztof Hałasa <[email protected]>
On bootup nvidiafb prints the following on my Apple G5:
nvidiafb: CRTC 1appears to have a CRT attached
There should be a space between the '1' and the 'appears'. Add it.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Kernel 2.6.30-rc1 added sys_preadv and sys_pwritev to most archs
but not ARM, resulting in
<stdin>:1421:2: warning: #warning syscall preadv not implemented
<stdin>:1425:2: warning: #warning syscall pwritev not implemented
This patch adds sys_preadv and sys_pwritev to ARM.
These syscalls simply take five long-sized parameters, so they
should have no calling-convention/ABI issues in the kernel.
Tested on armv5tel eabi using a preadv/pwritev test program posted
on linuxppc-dev earlier this month.
It would be nice to get this into the kernel before 2.6.30 final,
so that glibc's kernel version feature test for these syscalls
doesn't have to special-case ARM.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Russell King <[email protected]>
Compiling recent 2.6.29-rc kernels for ARM gives me the following warning:
arch/arm/mm/mmu.c: In function 'sanity_check_meminfo':
arch/arm/mm/mmu.c:697: warning: comparison between pointer and integer
This is because commit 3fd9825c42
"[ARM] 5402/1: fix a case of wrap-around in sanity_check_meminfo()"
in 2.6.29-rc5-git4 added a comparison of a pointer with PAGE_OFFSET,
which is an integer.
Fixed by casting PAGE_OFFSET to void *.
Signed-off-by: Mikael Pettersson <[email protected]>
Acked-by: Nicolas Pitre <[email protected]>
Signed-off-by: Russell King <[email protected]>
Promise ATA engines need to be reset when errors occur.
That's currently done for errors detected by sata_promise itself,
but it's not done for errors like timeouts detected outside of
the low-level driver.
The effect of this omission is that a timeout tends to result
in a sequence of failed COMRESETs after which libata EH gives
up and disables the port. At that point the port's ATA engine
hangs and even reloading the driver will not resume it.
To fix this, make sata_promise override ->hardreset on SATA
ports with code which calls pdc_reset_port() on the port in
question before calling libata's hardreset. PATA ports don't
use ->hardreset, so for those we override ->softreset instead.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
Impact: cleanup, no functionality changed
ia32_setup_rt_frame() has a duplicated code block labelled
"Make -mregparm=3 work" for setting up the register parameters
to the user-mode signal handler.
This is harmless but ugly. Remove the redundant assignments.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
Linux/ARM currently doesn't support robust or PI futexes.
The problem is that the kernel wants to perform certain ops
(cmpxchg, set, add, or, andn, xor) atomically on user-space
addresses, and ARM's futex.h doesn't support that.
This patch adds that support, but only for uniprocessor machines.
For UP it's enough to disable preemption to ensure mutual exclusion
with other software agents (futexes don't need to care about other
hardware agents, fortunately).
This patch is based on one posted by Khem Raj on 2007-08-01
<http://marc.info/?l=linux-arm-kernel&m=118599407413016&w=2>.
(That patch is included in the -RT kernel patches.)
My changes since that version include:
* corrected implementation of FUTEX_OP_ANDN (must complement oparg)
* added missing memory clobber to futex_atomic_cmpxchg_inatomic()
* removed spinlock because it's unnecessary for UP and insufficient
for SMP, instead the code is restricted to UP and relies on the
fact that pagefault_disable() also disables preemption
* coding style cleanups
Tested on ARMv5 XScales with the glibc-2.6 nptl test suite.
Tested-by: Bruce Ashfield <[email protected]>
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Russell King <[email protected]>
Building 2.6.27-rc1 on x86 with gcc-3.2.3 fails with:
In file included from include/asm/dma.h:12,
from include/linux/bootmem.h:8,
from init/main.c:26:
include/asm/io.h: In function `readb':
include/asm/io.h:32: syntax error before string constant
include/asm/io.h: In function `readw':
include/asm/io.h:33: syntax error before string constant
include/asm/io.h: In function `readl':
include/asm/io.h:34: syntax error before string constant
include/asm/io.h: In function `__readb':
include/asm/io.h:36: syntax error before string constant
include/asm/io.h: In function `__readw':
include/asm/io.h:37: syntax error before string constant
include/asm/io.h: In function `__readl':
include/asm/io.h:38: syntax error before string constant
make[1]: *** [init/main.o] Error 1
make: *** [init] Error 2
Starting with 2.6.27-rc1 readb() et al are generated by a
build_mmio_read() macro, which generates asm() statements with
output register constraints like "=" "q", i.e. as two adjacent
string literals. This doesn't work with gcc-3.2.3.
Fixed by moving the "=" part into the callers' reg parameter
(as suggested by Ingo).
Build and boot-tested with gcc-3.2.3 on 32 and 64-bit x86.
Fixes <http://bugzilla.kernel.org/show_bug.cgi?id=11205>.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
The arm ixp4xx_eth driver doesn't compile in 2.6.27-rc1:
CC [M] drivers/net/arm/ixp4xx_eth.o
drivers/net/arm/ixp4xx_eth.c: In function 'eth_poll':
drivers/net/arm/ixp4xx_eth.c:554: warning: passing argument 1 of 'dma_mapping_error' makes pointer from integer without a cast
drivers/net/arm/ixp4xx_eth.c:554: error: too few arguments to function 'dma_mapping_error'
drivers/net/arm/ixp4xx_eth.c: In function 'eth_xmit':
drivers/net/arm/ixp4xx_eth.c:701: warning: passing argument 1 of 'dma_mapping_error' makes pointer from integer without a cast
drivers/net/arm/ixp4xx_eth.c:701: error: too few arguments to function 'dma_mapping_error'
drivers/net/arm/ixp4xx_eth.c: In function 'init_queues':
drivers/net/arm/ixp4xx_eth.c:886: warning: passing argument 1 of 'dma_mapping_error' makes pointer from integer without a cast
drivers/net/arm/ixp4xx_eth.c:886: error: too few arguments to function 'dma_mapping_error'
make[3]: *** [drivers/net/arm/ixp4xx_eth.o] Error 1
make[2]: *** [drivers/net/arm] Error 2
make[1]: *** [drivers/net] Error 2
make: *** [drivers] Error 2
dma_mapping_error() changed in 2.6.27-rc1 to also take a device parameter,
but nobody bothered updating ixp4xx_eth.c. Fixed by passing the appropriate
device value in the dma_mapping_error() calls.
Tested on an ixp425 box.
Signed-off-by: Mikael Pettersson <[email protected]>
Acked-by: Krzysztof Halasa <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
Correct sparc64's implementation of FUTEX_OP_ANDN to do a
bitwise negate of the oparg parameter before applying the
AND operation. All other archs that support FUTEX_OP_ANDN
either negate oparg explicitly (frv, ia64, mips, sh, x86),
or do so indirectly by using an and-not instruction (powerpc).
Since sparc64 has and-not, I chose to use that solution.
I've not found any use of FUTEX_OP_ANDN in glibc so the
impact of this bug is probably minor. But other user-space
components may try to use it so it should still get fixed.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
Trying to build with CONFIG_NE2000=m fails with:
scripts/mod/modpost -o /tmp/tmp/linux-2.6.27-rc1/Module.symvers -S -s
ERROR: "NS8390_init" [drivers/net/ne.ko] undefined!
This is because the split of 8390 into pausing and non-pausing
versions was incompletely propagated to ne.c. This fixes it.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
Booting 2.6.26-rc6 on my 486 DX/4 fails with a "BUG: Int 6"
(invalid opcode) and a kernel halt immediately after the
kernel has been uncompressed. The BUG shows EIP pointing
to an rdtsc instruction in native_read_tsc(), invoked from
native_sched_clock().
(This error occurs so early that not even the serial console
can capture it.)
A bisection showed that this bug first occurs in 2.6.26-rc3-git7,
via commit 9ccc906c97:
>x86: distangle user disabled TSC from unstable
>
>tsc_enabled is set to 0 from the command line switch "notsc" and from
>the mark_tsc_unstable code. Seperate those functionalities and replace
>tsc_enable with tsc_disable. This makes also the native_sched_clock()
>decision when to use TSC understandable.
>
>Preparatory patch to solve the sched_clock() issue on 32 bit.
>
>Signed-off-by: Thomas Gleixner <[email protected]>
The core reason for this bug is that native_sched_clock() gets
called before tsc_init().
Before the commit above, tsc_32.c used a "tsc_enabled" variable
which defaulted to 0 == disabled, and which only got enabled late
in tsc_init(). Thus early calls to native_sched_clock() would skip
the TSC and use jiffies instead.
After the commit above, tsc_32.c uses a "tsc_disabled" variable
which defaults to 0, meaning that the TSC is Ok to use. Early calls
to native_sched_clock() now erroneously try to use the TSC on
!cpu_has_tsc processors, leading to invalid opcode exceptions.
My proposed fix is to initialise tsc_disabled to a "soft disabled"
state distinct from the hard disabled state set up by the "notsc"
kernel option. This fixes the native_sched_clock() problem. It also
allows tsc_init() to be simplified: instead of setting tsc_disabled = 1
on every error return, we just set tsc_disabled = 0 once when all
checks have succeeded.
I've verified that this lets my 486 boot again. I've also verified
that a Core2 machine still uses the TSC as clocksource after the patch.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
Minor coding-style fixes for sata_promise:
- remove stray blank lines
- fix checkpatch.pl errors; warnings about long lines
remain, but I don't intend to address those at this time
- remove two inline directives: neither is essential and
both functions are trivially inlinable anyway by virtue
of being static and having a single unique call site
- fix comment in pdc_interrupt(): the bits in PDC_INT_SEQMASK
denote SEQIDs not tags, the distinction becomes important
when NCQ gets implemented
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
This patch cleans up sata_promise's mmio accesses.
In sata_promise there are three distinct mmio address spaces:
1. global registers, offsets from host->iomap[PDC_MMIO_BAR]
2. per-port ATA registers, offsets from ap->ioaddr.cmd_addr
3. per-port SATA registers, offsets from ap->ioaddr.scr_addr
The driver currently often fails to indicate which address space
a given mmio base pointer refers to, which is a source of bugs
and confusion (see recent pdc_thaw() irq clearing bug; it's also
been an obstacle for the pending NCQ extensions).
To reduce these problems, adopt a coding style where the name of
a base pointer always indicates which address space it refers to:
1. global registers: host_mmio
2. per-port ATA registers: ata_mmio
3. per-port SATA registers: sata_mmio
Also rearrange register offset definitions to clearly indicate
which address space they belong to, and add a symbolic definition
for the previously hard-coded PHYMODE4 register.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
This patch fixes two bugs in sata_promise's irq status clearing paths:
1. When clearing the irq status for a specific port, the driver
read the global SEQMASK register. This is wrong because that
clears the irq status for _all_ ports.
2. pdc_thaw() incorrectly added the PDC_INT_SEQMASK host register
offset to a per-port ata engine base address. This resulted in
it reading the unrelated PDC_PKT_SUBMIT register, which did not
have the desired irq status clearing effect.
In both cases the fix is to read from the port's Command/Status
register. This also matches what Promise's own driver does.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
<asm-x86/nops.h> describes certain multibyte instructions as
"generic" nops when in fact they aren't nops at all in 64-bit
mode (missing REX.W causing truncation of a register).
Update the comment to state that K8 or P6 style nops should be
used in 64-bit mode. This matches what the alternatives code does.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
A Promise SATA controller will signal hotplug events when a hard
reset (COMRESET) is done on a port. These events aren't masked by
the driver, and the unexpected interrupts will cause a sequence
of failed reset attempts util libata's EH finally gives up.
This has not been a common problem so far, but the pending libata
hardreset-by-default changes makes it a critical issue.
The solution is to disable hotplug events before a reset, and to
reenable them afterwards. (Promise's driver does this too.)
This patch adds SATA-specific versions of ->freeze() and ->thaw()
that also disable and enable hotplug events. PATA ports continue
to use the old versions of ->freeze() and ->thaw().
Accesses to the hotplug register must be serialised via host->lock.
We rely on ap->lock == &ap->host->lock and that libata takes this
lock before ->freeze() and ->thaw(). Document this requirement.
The interrupt handler is adjusted so its hotplug register accesses
are inside the region protected by host->lock.
Tested on various chips (SATA300TX4, SATA300TX2plus, SATAII150TX4,
FastTrack TX4000) with various combinations of SATA and PATA disks,
with and without the pending hardreset-by-default changes.
Signed-off-by: Mikael Pettersson <[email protected]>
Acked-by: Tejun Heo <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
> Diffing dmesg between git7 and git8 doesn't sched any light since
> git8 also removed the printouts of the x86 caps as they were being
> initialised and updated. I'm currently adding those printouts back
> in the hope of seeing where and when the caps get broken.
That turned out to be very illuminating:
--- dmesg-2.6.24-git7 2008-02-24 18:01:25.295851000 +0100
+++ dmesg-2.6.24-git8 2008-02-24 18:01:25.530358000 +0100
...
CPU: After generic identify, caps: 00000003 00000000 00000000 00000000 00000000 00000000 00000000 00000000
CPU: After all inits, caps: 00000003 00000000 00000000 00000000 00000000 00000000 00000000 00000000
+CPU: After applying cleared_cpu_caps, caps: 00000013 00000000 00000000 00000000 00000000 00000000 00000000 00000000
Notice how the TSC cap bit goes from Off to On.
(The first two lines are printout loops from -git7 forward-ported
to -git8, the third line is the same printout loop added just after
the xor-with-cleared_cpu_caps[] loop.)
Here's how the breakage occurs:
1. arch/x86/kernel/tsc_32.c:tsc_init() sees !cpu_has_tsc,
so bails and calls setup_clear_cpu_cap(X86_FEATURE_TSC).
2. include/asm-x86/cpufeature.h:setup_clear_cpu_cap(bit) clears
the bit in boot_cpu_data and sets it in cleared_cpu_caps
3. arch/x86/kernel/cpu/common.c:identify_cpu() XORs all caps
in with cleared_cpu_caps
HOWEVER, at this point c->x86_capability correctly has TSC
Off, cleared_cpu_caps has TSC On, so the XOR incorrectly
sets TSC to On in c->x86_capability, with disastrous results.
The real bug is that clearing bits with XOR only works if the
bits are known to be 1 prior to the XOR, and that's not true here.
A simple fix is to convert the XOR to AND-NOT instead. The following
patch does that, and allows my 486 to boot 2.6.25-rc kernels again.
[ [email protected]: fixed a similar bug in setup_64.c as well. ]
The breakage was introduced via commit 7d851c8d3d.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
The original workaround for the Promise ASIC PRD bug
contained an endianess bug which I failed to detect:
the adjustment of the last PRD entry's length field
applied host arithmetic to little-endian data, which
is incorrect on big-endian machines.
We have the length available in host-endian format, so
do the adjustment on host-endian data and then convert
and store it in the PRD entry's little-endian data field.
Thanks to an anonymous reviewer for detecting this bug.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
Second-generation Promise SATA controllers have an ASIC bug
which can trigger if the last PRD entry is larger than 164 bytes,
resulting in intermittent errors and possible data corruption.
Work around this by replacing calls to ata_qc_prep() with a
private version that fills the PRD, checks the size of the
last entry, and if necessary splits it to avoid the bug.
Also reduce sg_tablesize by 1 to accommodate the new entry.
Tested on the second-generation SATA300 TX4 and SATA300 TX2plus,
and the first-generation PDC20378.
Thanks to Alexander Sabourenkov for verifying the bug by
studying the vendor driver, and for writing the initial patch
upon which this one is based.
Signed-off-by: Mikael Pettersson <[email protected]>
--
Changes since previous version:
* use new PDC_MAX_PRD constant to initialise sg_tablesize
drivers/ata/sata_promise.c | 87 ++++++++++++++++++++++++++++++++++++++++++---
1 files changed, 83 insertions(+), 4 deletions(-)
Signed-off-by: Jeff Garzik <[email protected]>
Fix a bitmask typo in the pdc202xx_new PLL frequency detection code
which causes it to truncate an intermediate difference to 26 bits
instead of the correct 30 bits (the PLL's bitwidth).
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Bartlomiej Zolnierkiewicz <[email protected]>
This patch corrects sata_promise to classify FastTrack TX4200
(DID 3515/3519) as a second-generation chip. Promise's partial-
source FT TX4200 driver confirms this classification.
Treating it as a first-generation chip causes several problems:
1. Detection failures. This is a recent regression triggered by
the hotplug-enabling changes in 2.6.23-rc1.
2. Various "failed to resume link for reset" warnings.
This patch fixes <http://bugzilla.kernel.org/show_bug.cgi?id=8936>.
Thanks to Stephen Ziemba for reporting the bug and for testing the fix.
Signed-off-by: Mikael Pettersson <[email protected]>
Tested-by: Stephen Ziemba <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
Previously I reported that the pata_pdc2027x PLL detection changes
in kernel 2.6.22 broke the driver on my PowerMac:
>pata_pdc2027x: Invalid PLL input clock 1691742kHz, give up!
This is followed by a number of errors and speed reduction
steps on the affected ports.
There are two bugs in pata_pdc2027x's PLL detection code:
1. The PLL counter's start value is read before the chip is
put in "test mode". Outside of test mode the counter is
halted, and on the PowerMac the counter is zero because
the chip hasn't been initialised by its BIOS.
The fix is to move the read of the start value to after
test mode is started, but before the mdelay() in test mode.
This also improves the precision of the PLL detection.
2. The code to compute the number of PLL decrements during the
mdelay() in test mode fails to consider that the PLL counter
only is 30 bits wide. If there is a wraparound, it will compute
an incorrect and much too large value. On the PowerMac, the
start count is zero, the end count is a large 30-bit value, so
wraparound occurs and an out of bounds PLL clock is detected.
The fix is to mask the (start - end) computation to 30 bits.
While debugging this I also noticed that pdc_read_counter()
reads the two halves of the 30-bit PLL counter as 16-bit values,
and then combines them as if the halves only are 15 bits wide.
To avoid confusion, the halves should be read as 15-bit values.
This patch implements all three changes. It fixes the PLL detection
failure on my PowerMac, and doesn't cause any regressions on an x86
with an identical card.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
Starting with kernel 2.6.23-rc1, the i386 APM driver fails
on several of my machines with the message:
apm: BIOS not found
This happens because of a bug in the i386 boot code rewrite
from assembler to C. The original assembly code had the
following code in its APM BIOS presence test (boot/setup.S):
andw $0x02, %cx # Is 32 bit supported?
je done_apm_bios # No 32-bit, no (good) APM BIOS
That is, the code bails out if bit 2 is zero.
In the new C version, this is coded as (boot/apm.c):
if (cx & 0x02) /* 32 bits supported? */
return -1;
Here we see that the test has been accidentally inverted.
The fix is to negate the test. I've verified that this
allows the APM driver to work again on my affected machines.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: H. Peter Anvin <[email protected]>
This patch enables hotplugging of SATA devices in the
sata_promise driver. It's been tested successfully on
both first- and second-generation Promise SATA chips:
SATA150 TX2plus, SATAII150 TX2plus, SATAII150 TX4,
SATA300 TX2plus, and SATA300 TX4.
The only quirk I've seen is that hotplugging (insertion)
on the first-generation SATA150 TX2plus requires a lengthier
EH sequence than on the second-generation chips.
On the second-generation chips a simple soft reset seems
to suffice, but on the first-generation chip there's a
"port is slow to respond" after the initial soft reset,
after which libata issues a hard reset, and then the
device is recognised.
The hotplug checks are high up in the interrupt handling
path, not deep down in error_intr as in ahci/sata_sil24.
That's because the chip doesn't signal hotplug status changes
in the per-port status register: instead a global register
contains hotplug control and status flags for all ports.
I considered following the ahci/sata_sil24 structure, but
that would have required non-trivial changes to the interrupt
handling path, so I chose to keep the hotplug changes simple
and unobtrusive.
Signed-off-by: Mikael Pettersson <[email protected]>
--
This patch depends on the "sata_promise: cleanups" patch.
Changes since the previous version (posted June 19):
- Correct pdc_interrupt() to increment 'handled' also in
the hotplug case. This prevents IRQ_NONE from being
returned when an interrupt only has hotplug events to
handle, which could confuse the kernel's IRQ machinery.
- Added testing on the SATAII150 TX4.
drivers/ata/sata_promise.c | 41 ++++++++++++++++++++++++++++++++++++-----
1 files changed, 36 insertions(+), 5 deletions(-)
Signed-off-by: Jeff Garzik <[email protected]>
sil24_interrupt() loads host->ports[i] into a local variable,
validates it, and then loads the value again in the call to
sil24_host_intr(). This patch replaces the second load by a
reference to the local variable.
This is safe since no side-effects have occurred since the
initial load. It also improves readability since it makes
it clear that the parameter to sil24_host_intr() is the same
value which was just validated.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
This patch applies some trivial cleanups to sata_promise:
- repair whitespace damage
- correct comment at board_2057x_pata definition
- pull SATAII TX4 support code out to separate functions
- rename ata_nr to ata_no for consistency with libata's port_no
- remove some init-time debug printks (requested by Jeff)
This patch should cause no behavioural changes, except for
the removed printks.
Signed-off-by: Mikael Pettersson <[email protected]>
--
drivers/ata/sata_promise.c | 56 ++++++++++++++++++---------------------------
1 files changed, 23 insertions(+), 33 deletions(-)
Signed-off-by: Jeff Garzik <[email protected]>
Compiling 2.6.22-rc1 with gcc-3.2.3 for i486 fails with:
gcc -m32 -Wp,-MD,net/core/.skbuff.o.d -nostdinc -isystem /home/mikpe/pkgs/linux-x86/gnu/lib/gcc-lib/i486-pc-linux-gnu/3.2.3/include -D__KERNEL__ -Iinclude -include include/linux/autoconf.h -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common -O2 -pipe -msoft-float -mregparm=3 -freg-struct-return -mpreferred-stack-boundary=4 -march=i486 -ffreestanding -maccumulate-outgoing-args -DCONFIG_AS_CFI=1 -Iinclude/asm-i386/mach-default -fomit-frame-pointer -D"KBUILD_STR(s)=#s" -D"KBUILD_BASENAME=KBUILD_STR(skbuff)" -D"KBUILD_MODNAME=KBUILD_STR(skbuff)" -c -o net/core/skbuff.o net/core/skbuff.c
net/core/skbuff.c:648:1: directives may not be used inside a macro argument
net/core/skbuff.c:647:39: unterminated argument list invoking macro "memcpy"
net/core/skbuff.c: In function `pskb_expand_head':
net/core/skbuff.c:651: `memcpy' undeclared (first use in this function)
net/core/skbuff.c:651: (Each undeclared identifier is reported only once
net/core/skbuff.c:651: for each function it appears in.)
net/core/skbuff.c:651: syntax error before "skb"
make[2]: *** [net/core/skbuff.o] Error 1
make[1]: *** [net/core] Error 2
make: *** [net] Error 2
The patch below implements a simple workaround which is to
clone the offending memcpy() call and specialise it for the
two different scenarios.
Other workarounds are of course possible: e.g. bind the varying
parameter in a local variable, or use a macro or inline function
to perform the varying computation.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
There is a known problem with sata_promise on SATAII-150/300 TX4
controller cards: it enumerates drives in an order that differs
from the port numbers printed on the controller cards. However,
Promise's BIOS and Linux driver both get the order right.
I investigated Promise's Linux driver (v1.01.0.23), and found
that it explicitly changes the mapping from logical port number
to ATA engine MMIO address on the SATAII TX4 cards. It does this
on all SATAII TX4 cards, without inspecting revision etc. The
SATAII TX2plus cards continue to use the same mapping that was
used for the first-generation chips.
This patch updates sata_promise to use the new port number to
ATA engine mapping on SATAII TX4 cards, which fixes the drive
enumeration order problem on those cards. Tested on several
1st and 2nd generation TX2plus and TX4 chips.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
The sata_promise error decode update changed pdc_host_intr()
to return and not complete the qc after detecting an error.
Unfortunately not completing the qc:s causes them to always
time out on error, which is wrong and has nasty side-effects.
This patch updates pdc_error_intr() to call ata_port_abort(),
similar to ahci and sata_sil24. Doing this is important as it
makes EH see the original error and not a bogus timeout.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
Promise ATA ports should always be reset by pdc_reset_port()
when errors are detected, but the recent error reason decoding
update to sata_promise replaced that reset with a freeze.
This patch changes the error detection to do a reset again.
This makes the error decoding update safer, as it now only
adds error decoding without changing any other behaviour.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
This patch adds much needed error reason decoding and
reporting to sata_promise. It's simplistic but should
log all relevant error info the controller provides.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
This patch changes sata_promise so that the PATA ports
on TX2plus chips are bound to the pdc_pata_ops structure.
This means that operations called from the SATA ops
structures don't need any SATA-vs-PATA tests any more.
Instead, operations that depend on a port being SATA or
PATA are separated into different procedures.
* pdc_cable_type() is split into a PATA version and a
SATA version
* pdc_error_handler() is split into a PATA version and a
SATA version, that both call a common version after
setting up the `hardreset' function pointer
* pdc_old_check_atapi_dma() is now only used for SATAI
ports, so is renamed to pdc_old_sata_check_atapi_dma()
and simplified
* pdc_sata_scr_{read,write}() are now only used for SATA
ports, so their is-not-SATA tests are removed
* pdc_port_start() is split into three procedures: a wrapper
which performs the ->ops adjustment on TX2plus PATA ports,
a procedure with the common code, and a procedure with
the SATA-specific code (this bit might be cleaned up by
Tejun's new init model)
Tested on 20619, 20575, and 20775 chips.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
The recent change which moved cable detection from
pdc_pre_reset() to the new ->cable_detect hook only
added the hook for SATAII chips, leaving SATAI chips
and the 20619 without the hook. Fixed by this patch.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
Compiling 2.6.21-rc5 with gcc-4.2.0 20070317 (prerelease)
for sparc64 fails as follows:
gcc -Wp,-MD,arch/sparc64/kernel/.time.o.d -nostdinc -isystem /home/mikpe/pkgs/linux-sparc64/gcc-4.2.0/lib/gcc/sparc64-unknown-linux-gnu/4.2.0/include -D__KERNEL__ -Iinclude -include include/linux/autoconf.h -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common -Os -m64 -pipe -mno-fpu -mcpu=ultrasparc -mcmodel=medlow -ffixed-g4 -ffixed-g5 -fcall-used-g7 -Wno-sign-compare -Wa,--undeclared-regs -fomit-frame-pointer -fno-stack-protector -Wdeclaration-after-statement -Wno-pointer-sign -Werror -D"KBUILD_STR(s)=#s" -D"KBUILD_BASENAME=KBUILD_STR(time)" -D"KBUILD_MODNAME=KBUILD_STR(time)" -c -o arch/sparc64/kernel/time.o arch/sparc64/kernel/time.c
cc1: warnings being treated as errors
arch/sparc64/kernel/time.c: In function 'kick_start_clock':
arch/sparc64/kernel/time.c:559: warning: overflow in implicit constant conversion
make[1]: *** [arch/sparc64/kernel/time.o] Error 1
make: *** [arch/sparc64/kernel] Error 2
gcc gets unhappy when the MSTK_SET macro's u8 __val variable
is updated with &= ~0xff (MSTK_YEAR_MASK). Making the constant
unsigned fixes the problem.
[ I fixed up the sparc32 side as well -DaveM ]
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
pata_legacy fails to detect the disk on my old ISA/VLB 486:
it starts to probe io=0x1f0 ctr=0x3f6 irq=15, complains
loudly about IDENTIFYs timing out, and finally fails.
(Sorry I couldn't capture the kernel's boot messages.)
It turns out that the driver's mapping from io to irq in
legacy_irq[] is wrong: index 0 for io=0x1f0 has irq=15 but
should have irq=14, and index 1 for io=0x170 has irq=14 but
should have irq=15. This is confirmed by a comparison with
include/asm-i386/ide.h:ide_default_irq().
This patch swaps the first two elements in legacy_irq[],
which makes pata_legacy work on my 486.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
Update my email address in CREDITS to one that works.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Each place in pdc_ata_init_one() that initialises a SATA
port first calls pdc_ata_setup_port(), and then manually
assigns the port's ->scr_addr. Simplify the code by extending
pdc_ata_setup_port() to also handle scr_addr initialisation;
for PATA ports we pass NULL as scr_addr.
The initialisation of the PATA-only 20619 redundantly set
up scr_addr for the ports. Remove this.
Tested on 20619, 20575, and 20775 chips.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
When I merged my 20619 new EH conversion with #libata-upstream
I had to manually resolve a conflict, and inadvertently lost
pdc_pata_ops' ->post_internal_cmd binding. Corrected by this patch.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
This patch updates the sata_promise driver to use new-style
libata error handling for 20619 (TX4000) chips. sata_promise
already uses new EH for the other chips it supports, so the
patch is quite simple:
* remove ->phy_reset and ->eng_timeout ops from pdc_pata_ops,
and instead bind ->freeze, ->thaw, ->error_handler, and
->post_internal_cmd to existing new EH functions
* drop ATA_FLAG_SRST from board_20619's flags
* remove now unused pdc_pata_phy_reset() and pdc_eng_timeout()
Tested on a TX4000 with both modern working disks and old/quirky
disks. Also used a CD-RW drive to test reading and writing CDs.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
This patch fixes an oversight which caused sata_promise to
not perform cable detection on the TX2plus chips' PATA ports.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
This patch extends sata_promise to handle ATAPI_NODATA
commands internally. However, commands destined to
ATA_DFLAG_CDB_INTR devices are excluded from this and
continue to be returned to libata.
Concrete changes:
- pdc_atapi_dma_pkt() is renamed to pdc_atapi_pkt(), and is
extended to set up correct headers for NODATA packets
- pdc_qc_prep() calls pdc_atapi_pkt() for ATAPI_NODATA
- pdc_host_intr() handles ATAPI_NODATA
- pdc_qc_issue_prot() sends ATAPI_NODATA packets via the
chip's packet mechanism, except for CDB_INTR devices
Tested on first- and second-generation chips, SATAPI and PATAPI,
with no observable regressions.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
This patch (against libata #upstream + the ATAPI cleanup patch)
reimplements sata_promise's ATAPI support to format ATAPI DMA
commands as normal packets, and to issue them via the hardware's
normal packet machinery.
It turns out that the only reason for issuing ATAPI DMA
commands via the pdc_issue_atapi_pkt_cmd() procedure was to
perform two interrupt-fiddling steps for ATA_DFLAG_CDB_INTR
devices. But these steps aren't needed because sata_promise
sets ATA_FLAG_PIO_POLLING, which disables DMA for those devices.
The remaining steps can easily be done in ATA taskfile packets.
Concrete changes:
- pdc_atapi_dma_pkt() is extended to program all packet setup
steps, and not just contain the CDB; the sequence of steps
exactly mirrors what pdc_issue_atapi_pkt_cmd() did
- pdc_atapi_dma_pkt() needed more parameters: simplify it by
just passing 'qc' and having it extract the data it needs
- pdc_issue_atai_pkt_cmd() and its two helper procedures
pdc_wait_for_drq() and pdc_wait_on_busy() are removed
Tested on first- and second-generation chips, SATAPI and PATAPI,
with no observable regressions.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
Here's a cleanup for yesterday's sata_promise ATAPI patch:
- add and use a symbolic constant for the altstatus register
- check return status from ata_busy_wait()
- add missing newline in a warning printk()
- update comment in pdc_issue_atapi_pkt_cmd() to clarify
that the maybe-wait-for-INT issue cannot occur in the
current driver, but may occur if the driver starts issuing
ATAPI non-DMA commands as PDC packets
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
This patch adds ATAPI support to the sata_promise driver.
This has been tested on both first- and second-generation
chips (20378 and 20575), and with both SATAPI and PATAPI
devices. CD-writing works.
SATAPI DMA works on second-generation chips, but on
first-generation chips SATAPI is limited to PIO due
to what appears to be HW limitations.
PATAPI DMA works on both first- and second-generation
chips, but requires the separate PATA support patch
before it can be used on TX2plus chips.
The functional changes to the driver are:
- remove ATA_FLAG_NO_ATAPI from PDC_COMMON_FLAGS
- add ->check_atapi_dma() operation to enable DMA for bulk data
transfers but force PIO for other ATAPI commands; this filter
is from Promise's driver and largely matches pata_pdc207x.c
- use a more restrictive ->check_atapi_dma() on first-generation
chips to force SATAPI to always use PIO
- add handling of ATAPI protocols to pdc_qc_prep(), pdc_host_intr(),
and pdc_qc_issue_prot(): ATAPI_DMA is handled by the driver
while non-DMA protocols are handed over to libata generic code
- add pdc_issue_atapi_pkt_cmd() to handle the initial steps in
issuing ATAPI DMA commands before sending the actual CDB;
this procedure was ported from Promise's driver
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
This patch implements a simple way of setting up per-port
flags on the SATA+PATA Promise TX2plus chips, which is a
prerequisite for supporting the PATA port on those chips.
It is based on the observation that ap->flags isn't really
used until after ->port_start() has been invoked. So it
places the "exceptional" per-port flags array in the driver's
private host structure, and uses it in ->port_start() to
finalise the port's flags.
This patch obsoletes the #promise-sata-pata branch included
in the #all branch.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
In file included from net/netfilter/xt_state.c:13:
include/net/netfilter/nf_conntrack_compat.h: In function 'nf_ct_l3proto_try_module_get':
include/net/netfilter/nf_conntrack_compat.h:70: error: 'PF_INET' undeclared (first use in this function)
include/net/netfilter/nf_conntrack_compat.h:70: error: (Each undeclared identifier is reported only once
include/net/netfilter/nf_conntrack_compat.h:70: error: for each function it appears in.)
include/net/netfilter/nf_conntrack_compat.h:71: warning: control reaches end of non-void function
make[2]: *** [net/netfilter/xt_state.o] Error 1
make[1]: *** [net/netfilter] Error 2
make: *** [net] Error 2
A simple fix is to have nf_conntrack_compat.h #include <linux/socket.h>.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Patrick McHardy <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
This patch adds myself as maintainer of the sata_promise
libata driver.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
include/linux/utsrelease.h and include/linux/version.h aren't removed any
more by mrproper in kernel 2.6.20-rc2. The patch below fixes this.
The definition of MRPROPER_FILES looks weird: generated-headers looks like
a misspelling of generated_headers, but that one is a Makefile target, not
a variable or a file, so I don't see how including it in MRPROPER_FILES
could have any effect.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
This patch converts sata_promise to use new-style libata error
handling on Promise SATA chips, for both SATA and PATA ports.
* ATA_FLAG_SRST is no longer set
* ->phy_reset is no longer set as it is unused when ->error_handler
is present, and pdc_sata_phy_reset() has been removed
* pdc_freeze() masks interrupts and halts DMA via PDC_CTLSTAT
* pdc_thaw() clears interrupt status in PDC_INT_SEQMASK and then
unmasks interrupts in PDC_CTLSTAT
* pdc_error_handler() reinitialises the port if it isn't frozen,
and then invokes ata_do_eh() with standard {s,}ata reset methods
* pdc_post_internal_cmd() resets the port in case of errors
* the PATA-only 20619 chip continues to use old-style EH:
not by necessity but simply because I don't have documentation
for it or any way to test it
Since the previous version pdc_error_handler() has been rewritten
and it now mostly matches ahci and sata_sil24. In case anyone
wonders: the call to pdc_reset_port() isn't a heavy-duty reset,
it's a light-weight reset to quickly put a port into a sane state.
The discussion about the PCI flushes in pdc_freeze() and pdc_thaw()
seemed to end with a consensus that the flushes are OK and not
obviously redundant, so I decided to keep them for now.
This patch was prepared against 2.6.19-git7, but it also applies
to 2.6.19 + libata #upstream, with or without the revised sata_promise
cleanup patch I recently submitted.
This patch does conflict with the #promise-sata-pata patch:
this patch removes pdc_sata_phy_reset() while #promise-sata-pata
modifies it. The correct patch resolution is to remove the function.
Tested on 2037x and 2057x chips, with PATA patches on top and disks
on both SATA and PATA ports.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
This patch performs two simple cleanups of sata_promise.
* Remove board_20771 and map device id 0x3577 to board_2057x.
After the recent corrections for SATAII chips, board_20771 and
board_2057x were equivalent in the driver.
* Remove hp->hotplug_offset and use hp->flags & PDC_FLAG_GEN_II
to compute hotplug_offset in pdc_host_init(). hp->hotplug_offset
was used to distinguish 1st and 2nd generation chips in one
particular case, but now we have that information in a more
general form in hp->flags, so hp->hotplug_offset is redundant.
Changes since previous submission: rebased on libata-dev #upstream,
cleaned up hotplug_offset computation based on Tejun's comments,
expanded hotplug_offset removal rationale.
This patch does not depend on the pending new EH conversion patch.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
This patch adds code to fix up the PHYMODE4 "align timing"
register value on second-generation Promise SATA chips.
Failure to correct this value on non-x86 machines makes
drive detection prone to failure due to timeouts. (I've
observed about 50% detection failure rates on SPARC64.)
The HW boots with a bad value in this register, but on x86
machines the Promise BIOS corrects it to the value recommended
by the manual, so most people have been unaffected by this issue.
After developing the patch I checked Promise's SATAII driver,
and discovered that it also corrects PHYMODE4 just like this
patch does.
This patch depends on the sata_promise SATAII updates
patch I sent recently.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
This patch updates the sata_promise driver as follows:
- Correct typo in definition of PDC_TBG_MODE: it's at 0x41C not 0x41
in first-generation chips. This error caused PCI access alignment
exceptions on SPARC64, and on all platforms it disabled the expected
initialisation of TBG mode.
- Add flags field to struct pdc_host_priv. Define PDC_FLAG_GEN_II
and use it to distinguish first- and second-generation chips.
- Prevent the FLASH_CTL FIFO_SHD bit from being set to 1 on second-
generation chips. This matches Promises' ulsata2 driver.
- Prevent TBG mode and SLEW rate initialisation in second-generation chips.
These two registers have moved, TBG mode has been redefined, and
Promise's ulsata2 driver no longer attempts to initialise them.
- Correct PCI device table so devices 0x3570, 0x3571, and 0x3d73 are
marked as 2057x (2nd gen) not 2037x (1st gen).
- Correct PCI device table so device 0x3d17 is marked as 40518
(2nd gen 4 ports) not 20319 (1st gen 4 ports).
- Correct pdc_ata_init_one() to treat 20771 as a second-generation chip.
Tested on 0x3d75 (2nd gen), 0x3d73 (2nd gen), and 0x3373 (1st gen) chips.
The information comes from the newly uploaded Promise SATA HW specs,
Promise's ultra and ulsata2 drivers, and debugging on 3d75/3d73/3373 chips.
hp->hotplug_offset could now be removed and its value recomputed
in pdc_host_init() using hp->flags, but that would be a cleanup
not a functional change, so I'm ignoring it for now.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: Jeff Garzik <[email protected]>
Implementations assume the buffer is at least 4 byte aligned.
Signed-off-by: David S. Miller <[email protected]>
Cc: David Brownell <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>
The sparc64 kernel's EXPORT_SYMBOL(_mcount) is inside an
#ifdef CONFIG_SMP. This breaks modules in non-SMP kernels
built with stack overflow checking (CONFIG_STACK_DEBUG=y),
as modules_install reports:
WARNING: /lib/modules/2.6.17/kernel/drivers/ide/ide-cd.ko needs unknown symbol _mcount
Trivially fixed by moving EXPORT_SYMBOL(_mcount) outside of
the #ifdef CONFIG_SMP.
Signed-off-by: Mikael Pettersson <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
The PC Speaker driver's ->probe() routine doesn't even get called in the
64-bit kernels. The reason for that is that the arch code apparently has
to explictly add a "pcspkr" platform device in order for the driver core to
call the ->probe() routine. arch/i386/kernel/setup.c unconditionally adds
a "pcspkr" device, but the x86_64 kernel has no code at all related to the
PC Speaker.
The patch below copies the relevant code from i386 to x86_64, which makes
the PC Speaker work for me on x86_64.
Cc: Dmitry Torokhov <[email protected]>
Acked-by: Andi Kleen <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Mikael Pettersson