Messages in this thread |  | | Date | Tue, 8 Jan 2013 18:49:51 +0100 | From | Andrea Arcangeli <> | Subject | Re: oops in copy_page_rep() |
| |
Hi Kirill,
On Tue, Jan 08, 2013 at 07:30:58PM +0200, Kirill A. Shutemov wrote: > Merged patch is obviously broken: huge_pmd_set_accessed() can be called > only if the pmd is under splitting.
Of course I assume you meant "only if the pmd is not under splitting".
But no, setting a bitflag like the young bit or clearing or setting the numa bit won't screw with split_huge_page and it's safe even if the pmd is under splitting.
Those bits are only checked here at the last stage of split_huge_page_map after taking the PT lock:
spin_lock(&mm->page_table_lock); pmd = page_check_address_pmd(page, mm, address, PAGE_CHECK_ADDRESS_PMD_SPLITTING_FLAG); if (pmd) { pgtable = pgtable_trans_huge_withdraw(mm); pmd_populate(mm, &_pmd, pgtable);
haddr = address; for (i = 0; i < HPAGE_PMD_NR; i++, haddr += PAGE_SIZE) { pte_t *pte, entry; BUG_ON(PageCompound(page+i)); entry = mk_pte(page + i, vma->vm_page_prot); entry = maybe_mkwrite(pte_mkdirty(entry), vma); if (!pmd_write(*pmd)) entry = pte_wrprotect(entry); else BUG_ON(page_mapcount(page) != 1); if (!pmd_young(*pmd)) entry = pte_mkold(entry); if (pmd_numa(*pmd)) entry = pte_mknuma(entry); pte = pte_offset_map(&_pmd, haddr); BUG_ON(!pte_none(*pte)); set_pte_at(mm, haddr, pte, entry); pte_unmap(pte); }
If "young" or "numa" bitflags changed on the original *pmd for the previous part of split_huge_page, nothing will go wrong by the time we get to split_huge_page_map (the same is not true if the pfn changes!).
If you think this is too tricky, we could also decide to forbid huge_pmd_set_accessed if the pmd is in splitting state, but I don't think that flipping young/numa bits while in splitting state, can cause any problem (if done correctly with PT lock + pmd_same).
Thanks!
|  |