From 0106d456c4cb1770253fefc0ab23c9ca760b43f7 Mon Sep 17 00:00:00 2001 From: Will Deacon Date: Tue, 7 Jun 2016 17:55:15 +0100 Subject: arm64: mm: always take dirty state from new pte in ptep_set_access_flags Commit 66dbd6e61a52 ("arm64: Implement ptep_set_access_flags() for hardware AF/DBM") ensured that pte flags are updated atomically in the face of potential concurrent, hardware-assisted updates. However, Alex reports that: | This patch breaks swapping for me. | In the broken case, you'll see either systemd cpu time spike (because | it's stuck in a page fault loop) or the system hang (because the | application owning the screen is stuck in a page fault loop). It turns out that this is because the 'dirty' argument to ptep_set_access_flags is always 0 for read faults, and so we can't use it to set PTE_RDONLY. The failing sequence is: 1. We put down a PTE_WRITE | PTE_DIRTY | PTE_AF pte 2. Memory pressure -> pte_mkold(pte) -> clear PTE_AF 3. A read faults due to the missing access flag 4. ptep_set_access_flags is called with dirty = 0, due to the read fault 5. pte is then made PTE_WRITE | PTE_DIRTY | PTE_AF | PTE_RDONLY (!) 6. A write faults, but pte_write is true so we get stuck The solution is to check the new page table entry (as would be done by the generic, non-atomic definition of ptep_set_access_flags that just calls set_pte_at) to establish the dirty state. Cc: # 4.3+ Fixes: 66dbd6e61a52 ("arm64: Implement ptep_set_access_flags() for hardware AF/DBM") Reviewed-by: Catalin Marinas Reported-by: Alexander Graf Tested-by: Alexander Graf Signed-off-by: Will Deacon --- arch/arm64/mm/fault.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/arm64/mm/fault.c') diff --git a/arch/arm64/mm/fault.c b/arch/arm64/mm/fault.c index 5954881a35ac..ba3fc12bd272 100644 --- a/arch/arm64/mm/fault.c +++ b/arch/arm64/mm/fault.c @@ -109,7 +109,7 @@ int ptep_set_access_flags(struct vm_area_struct *vma, * PTE_RDONLY is cleared by default in the asm below, so set it in * back if necessary (read-only or clean PTE). */ - if (!pte_write(entry) || !dirty) + if (!pte_write(entry) || !pte_sw_dirty(entry)) pte_val(entry) |= PTE_RDONLY; /* -- cgit From bbb1681ee3653bdcfc6a4ba31902738118311fd4 Mon Sep 17 00:00:00 2001 From: Mark Rutland Date: Mon, 13 Jun 2016 17:57:02 +0100 Subject: arm64: mm: mark fault_info table const Unlike the debug_fault_info table, we never intentionally alter the fault_info table at runtime, and all derived pointers are treated as const currently. Make the table const so that it can be placed in .rodata and protected from unintentional writes, as we do for the syscall tables. Signed-off-by: Mark Rutland Cc: Catalin Marinas Cc: Will Deacon Signed-off-by: Will Deacon --- arch/arm64/mm/fault.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/arm64/mm/fault.c') diff --git a/arch/arm64/mm/fault.c b/arch/arm64/mm/fault.c index ba3fc12bd272..013e2cbe7924 100644 --- a/arch/arm64/mm/fault.c +++ b/arch/arm64/mm/fault.c @@ -441,7 +441,7 @@ static int do_bad(unsigned long addr, unsigned int esr, struct pt_regs *regs) return 1; } -static struct fault_info { +static const struct fault_info { int (*fn)(unsigned long addr, unsigned int esr, struct pt_regs *regs); int sig; int code; -- cgit