Hey,
2.6.26-rc2 with fixed vmlinux.lds netboots and with NFS root seems
to mostly work. But it did seem to deadlock. From the TOC (PIM) dump,
looks like a lock didn't get unlocked and both CPUs where spinning
trying to acquire that lock:
CPU0 IP : 0x4011a23c _spin_lock+14
CPU0 GR02: 0x4011cb6c flush_kernel_dcache_page_addr+34
CPU1 IP : 0x4011a23c _spin_lock+14
CPU1 GR02: 0x4011cabc __flush_tlb_range+74
flush_kernel_dcache_page_addr() is pretty short:
void flush_kernel_dcache_page_addr(void *addr)
{
flush_kernel_dcache_page_asm(addr);
purge_tlb_start();
pdtlb_kernel(addr);
purge_tlb_end();
}
Since I just updated my tree to 2.6.26-rc3, I'll use the asm from that:
...
20: 2b 60 00 00 addil L%0,dp,r1
20: R_PARISC_DLTIND21L pa_tlb_lock
24: 50 3a 00 00 ldd 0(r1),r26
24: R_PARISC_DLTIND14R pa_tlb_lock
28: 37 dd 3f a1 ldo -30(sp),ret1
2c: e8 00 a0 00 b,l 34 <flush_kernel_dcache_page_addr+0x34>,rp
2c: R_PARISC_PCREL22F _spin_lock
30: 08 1b 02 44 copy dp,r4
34: 08 04 02 5b copy r4,dp
38: 04 a0 12 00 pdtlb r0(r5)
3c: 34 1c 00 02 ldi 1,ret0
...
So while the offset looks wrong, __flush_tlb_range has a similar
construct:
5c: 2b 60 00 00 addil L%0,dp,r1
5c: R_PARISC_DLTIND21L pa_tlb_lock
60: 08 1b 02 44 copy dp,r4
64: 50 26 00 00 ldd 0(r1),r6
64: R_PARISC_DLTIND14R pa_tlb_lock
68: 37 dd 3f a1 ldo -30(sp),ret1
6c: e8 00 a0 00 b,l 74 <__flush_tlb_range+0x74>,rp
6c: R_PARISC_PCREL22F _spin_lock
70: 08 06 02 5a copy r6,r26
74: 08 04 02 5b copy r4,dp
78: 2b 60 00 00 addil L%0,dp,r1
78: R_PARISC_DLTIND21L split_tlb
7c: 50 3c 00 00 ldd 0(r1),ret0
7c: R_PARISC_DLTIND14R split_tlb
80: 0f 80 10 9f ldw 0(ret0),r31
84: 87 e0 20 72 cmpib,=,n 0,r31,c4 <__flush_tlb_range+0xc4>
88: 9c a0 20 28 cmpb,*= r0,r5,a4 <__flush_tlb_range+0xa4>
So I'm pretty sure, whatever is wrong, it's got something to do
with pa_tlb_lock.
Rebuilding now with gcc-4.2 (instead of gcc-4.3) and retesting
with 2.6.26-rc3.
hth,
grant
--
To unsubscribe from this list: send the line "unsubscribe linux-parisc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
