kernel.git/arch/sh, branch linux-2.6.21.y

sh: Kill bogus GCC4 symbol exports.

2007-03-28T08:24:47Z

__sdivsi3_i4i, __udiv_qrnnd_16, and __udivsi3_i4i don't exist outside of the ST compiler, so kill them off. This causes compile failures with other GCC4 compilers. Signed-off-by: Paul Mundt

sh: Fix bogus regs pointer in do_IRQ().

2007-03-14T04:03:35Z

SH-3 and SH-4 were trampling the register, and SH-2 wasn't even setting it in the first place. This ended up with some rather broken behaviour in the sysrq show_regs(). Signed-off-by: Paul Mundt

sh: Fix SH-3 cache entry_mask and way_size calculation.

2007-03-12T07:15:22Z

The code for performing the calculation was only in the SH-4 probe path, move it out to the common path so the other parts get this right too. Signed-off-by: Paul Mundt

sh: Fix PCI BAR address-space wraparound.

2007-03-12T05:50:49Z

When a SH7751R system includes a card that has wide range space like a graphics card, the pci-pci bridge controller can't set the correct address range. For example, when *lower_limit is 0xfd000000 and bar_size is 0x4000000, in the following code at arch/sh/drivers/pci/pci-auto.c, 0x0 is set in bar_value. pciauto_setup_bars() { ... bar_value = ((*lower_limit - 1) & ~(bar_size - 1)) + bar_size; ... *lower_limit = bar_value + bar_size; } As a result, 0x4000000 is set in *lower_limit, but this value is wrong. The following patch avoids this problem by checking the range of the value and refusing to update the BAR if the calculated value ends up being bogus. Signed-off-by: Hideo Saito Signed-off-by: Paul Mundt

sh: Kill off I/O cruft for R7780RP.

2007-03-05T11:19:57Z

We don't have any use for these machvec fixups anymore, kill them all off and go with the generic instead. Signed-off-by: Paul Mundt

sh: Revert lazy dcache writeback changes.

2007-03-05T10:46:47Z

These ended up causing too many problems on older parts, revert for now.. Signed-off-by: Paul Mundt

sh: Enable SM501 support for RTS7751R2D.

2007-03-05T05:13:26Z

This enables the SM501 drivers for the R2D board. Additional work needs to be done to migrate off of the VoyagerGX cchip code to make use of the rest of the mfd infrastructure. Signed-off-by: Paul Mundt

sh: Use L1_CACHE_BYTES for .data.cacheline_aligned.

2007-03-05T05:13:26Z

Previously this was using a hardcoded 32, use L1_CACHE_BYTES for cacheline alignment instead. Signed-off-by: Paul Mundt

sh: Fix kernel thread stack corruption with preempt.

2007-03-05T05:13:26Z

When I run a preemptive kernel-2.6.20 for SH7780, a created kthread(pdflush) can not exit by do_exit() in kernel_thread_helper. I think that the created kthread should have a room for 'struct pt_regs' space on the stack top, because __switch_to() will refer to the space as follows using 'regs = task_pt_regs(prev)' and next condition may be true. Signed-off-by: Hideo Saito Signed-off-by: Paul Mundt

sh: Fix sigmask trampling in signal delivery.

2007-03-05T05:13:25Z

There was a missing return in do_signal() that caused the saved sigmask to be written back after having successfully delivered the signal. Signed-off-by: Ryusuke Sakato Signed-off-by: Paul Mundt