We are still troubled by occasional random segmentation faults and memory memory corruption on SMP machines. The causes quite a few package builds to fail on the Debian buildd machines for parisc. When gcc-6 failed to build three times in a row, I looked again at the TLB related code. I found a couple of issues. This is the first. In general, we need to ensure page table updates and corresponding TLB purges are atomic. The attached patch fixes an instance in pci-dma.c where the page table update was not guarded by the TLB lock. Tested on rp3440 and c8000. So far, no further random segmentation faults have been observed.
Unified · Split
| @@ -95,8 +95,8 @@ static inline int map_pte_uncached(pte_t * pte, · 94 unmodified lines | |||
| 95 | 95 | ||
| 96 | if (!pte_none(*pte)) | 96 | if (!pte_none(*pte)) |
| 97 | printk(KERN_ERR "map_pte_uncached: page already exists\n"); | 97 | printk(KERN_ERR "map_pte_uncached: page already exists\n"); |
| 98 | set_pte(pte, __mk_pte(*paddr_ptr, PAGE_KERNEL_UNC)); | ||
| 99 | purge_tlb_start(flags); | 98 | purge_tlb_start(flags); |
| 99 | set_pte(pte, __mk_pte(*paddr_ptr, PAGE_KERNEL_UNC)); | ||
| 100 | pdtlb_kernel(orig_vaddr); | 100 | pdtlb_kernel(orig_vaddr); |
| 101 | purge_tlb_end(flags); | 101 | purge_tlb_end(flags); |
| 102 | vaddr += PAGE_SIZE; | 102 | vaddr += PAGE_SIZE; |