kernel.git/arch/mips/mm, branch linux-3.17.y

MIPS: tlbex: Fix potential HTW race on TLBL/M/S handlers

2014-12-06T23:57:14Z

commit 070e76cb3ffe43f6855492e77c96680c562598f0 upstream. There is a potential race when probing the TLB in TLBL/M/S exception handlers for a matching entry. Between the time we hit a TLBL/S/M exception and the time we get to execute the TLBP instruction, the HTW may have replaced the TLB entry we are interested in hence the TLB probe may fail. However, in the existing handlers, we never checked the status of the TLBP (ie check the result in the C0/Index register). We fix this by adding such a check when the core implements the HTW. If we couldn't find a matching entry, we return back and try again. Signed-off-by: Leonid Yegoshin Signed-off-by: Markos Chandras Reviewed-by: James Hogan Cc: linux-mips@linux-mips.org Patchwork: https://patchwork.linux-mips.org/patch/8599/ Signed-off-by: Ralf Baechle Signed-off-by: Greg Kroah-Hartman

MIPS: tlb-r4k: Add missing HTW stop/start sequences

2014-12-06T23:57:12Z

commit 6a8dff6ab16c903b0d8ef5fbf21543f39bf5d675 upstream. HTW needs to stop and start again whenever the EntryHI register changes otherwise an inflight HTW operation might use the new EntryHI register for updating an old entry and that could lead to crashes or even a machine check exception. We fix this by ensuring the HTW has stop whenever the EntryHI register is about to change Signed-off-by: Markos Chandras Cc: linux-mips@linux-mips.org Patchwork: https://patchwork.linux-mips.org/patch/8511/ Signed-off-by: Ralf Baechle Signed-off-by: Greg Kroah-Hartman

MIPS: tlbex: Properly fix HUGE TLB Refill exception handler

2014-11-14T18:10:29Z

commit 9e0f162a36914937a937358fcb45e0609ef2bfc4 upstream. In commit 8393c524a25609 (MIPS: tlbex: Fix a missing statement for HUGETLB), the TLB Refill handler was fixed so that non-OCTEON targets would work properly with huge pages. The change was incorrect in that it broke the OCTEON case. The problem is shown here: xxx0: df7a0000 ld k0,0(k1) . . . xxxc0: df610000 ld at,0(k1) xxxc4: 335a0ff0 andi k0,k0,0xff0 xxxc8: e825ffcd bbit1 at,0x5,0x0 xxxcc: 003ad82d daddu k1,at,k0 . . . In the non-octeon case there is a destructive test for the huge PTE bit, and then at 0, $k0 is reloaded (that is what the 8393c524a25609 patch added). In the octeon case, we modify k1 in the branch delay slot, but we never need k0 again, so the new load is not needed, but since k1 is modified, if we do the load, we load from a garbage location and then get a nested TLB Refill, which is seen in userspace as either SIGBUS or SIGSEGV (depending on the garbage). The real fix is to only do this reloading if it is needed, and never where it is harmful. Signed-off-by: David Daney Cc: Huacai Chen Cc: Fuxin Zhang Cc: Zhangjin Wu Cc: linux-mips@linux-mips.org Patchwork: https://patchwork.linux-mips.org/patch/8151/ Signed-off-by: Ralf Baechle Signed-off-by: Greg Kroah-Hartman

Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm

2014-09-22T18:58:23Z

Pull KVM fixes from Paolo Bonzini: "Two very simple bugfixes, affecting all supported architectures" [ Two? There's three commits in here. Oh well, I guess Paolo didn't count the preparatory symbol export ] * tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: KVM: correct null pid check in kvm_vcpu_yield_to() KVM: check for !is_zero_pfn() in kvm_is_mmio_pfn() mm: export symbol dependencies of is_zero_pfn()

mm: export symbol dependencies of is_zero_pfn()

2014-09-14T14:25:14Z

In order to make the static inline function is_zero_pfn() callable by modules, export its symbol dependencies 'zero_pfn' and (for s390 and mips) 'zero_page_mask'. We need this for KVM, as CONFIG_KVM is a tristate for all supported architectures except ARM and arm64, and testing a pfn whether it refers to the zero page is required to correctly distinguish the zero page from other special RAM ranges that may also have the PG_reserved bit set, but need to be treated as MMIO memory. Signed-off-by: Ard Biesheuvel Acked-by: Andrew Morton Signed-off-by: Paolo Bonzini

MIPS: Remove race window in page fault handling

2014-08-19T11:31:09Z

Multicore MIPSes without I/D hardware coherency suffered from a race condition in the page fault handler. The page table entry was published before any pending lazy D-cache flush was committed, hence it allowed execution of stale page cache data by other VPEs in the system. To make the cache handling safe we need to perform flushing already in the set_pte_at function. MIPSes without coherent I-caches can get a small increase in flushes due to the unavailability of the execute flag in set_pte_at. [ralf@linux-mips.org: outlining set_pte_at() saves a good k in a test build, so I moved its definition from pgtable.h to cache.c.] Signed-off-by: Lars Persson Cc: linux-mips@linux-mips.org Patchwork: https://patchwork.linux-mips.org/patch/7511/ Signed-off-by: Ralf Baechle

Merge branch '3.16-fixes' into mips-for-linux-next

2014-08-01T22:07:03Z

MIPS: Initialise MAARs

2014-08-01T22:06:46Z

Add initialisation for Memory Accessibility Attribute Registers. Generic code cannot know the platform-specific requirements with regards to speculative accesses, so it simply calls a platform_maar_init function which platforms with MAARs are expected to implement by calling the provided write_maar_pair function & returning the number of MAAR pairs used. A weak default implementation will simply use no MAAR pairs. Any present but unused MAAR pairs are then marked invalid, effectively disabling them. The end result of this patch is that MAARs are all marked invalid, until platforms implement the platform_maar_init function. Signed-off-by: Paul Burton Cc: linux-mips@linux-mips.org Patchwork: https://patchwork.linux-mips.org/patch/7331/ Signed-off-by: Ralf Baechle

MIPS: Use dedicated exception handler if CPU supports RI/XI exceptions

2014-08-01T22:06:39Z

Use the regular tlb_do_page_fault_0 (no write) handler to handle the RI and XI exceptions. Also skip the RI/XI validation check on TLB load handler since it's redundant when the CPU has unique RI/XI exceptions. Singed-off-by: Leonid Yegoshin Signed-off-by: Markos Chandras Cc: linux-mips@linux-mips.org Patchwork: https://patchwork.linux-mips.org/patch/7339/ Signed-off-by: Ralf Baechle

MIPS: mm: Use the Hardware Page Table Walker if the core supports it

2014-08-01T22:06:39Z

The Hardware Page Table Walker aims to speed up TLB refill exceptions by handling them in the hardware level instead of having a software TLB refill handler. However, a TLB refill exception can still be thrown in certain cases such as, synchronus exceptions, or address translation or memory errors during the HTW operation. As a result of which, HTW must not be considered a complete replacement for the TLB refill software handler, but rather a fast-path for it. For HTW to work, the PWBase register must contain the task's page global directory address so the HTW will kick in on TLB refill exceptions. Due to HTW being a separate engine embedded deep in the CPU pipeline, we need to restart the HTW everytime a PTE changes to avoid HTW fetching a old entry from the page tables. It's also necessary to restart the HTW on context switches to prevent it from fetching a page from the previous process. Finally, since HTW is using the entryhi register to write the translations to the TLB, it's necessary to stop the HTW whenever the entryhi changes (eg for tlb probe perations) and enable it back afterwards. == Performance == The following trivial test was used to measure the performance of the HTW. Using the same root filesystem, the following command was used to measure the number of tlb refill handler executions with and without (using 'nohtw' kernel parameter) HTW support. The kernel was modified to use a scratch register as a counter for the TLB refill exceptions. find /usr -type f -exec ls -lh {} \; HTW Enabled: TLB refill exceptions: 12306 HTW Disabled: TLB refill exceptions: 17805 Signed-off-by: Markos Chandras Cc: linux-mips@linux-mips.org Cc: Markos Chandras Patchwork: https://patchwork.linux-mips.org/patch/7336/ Signed-off-by: Ralf Baechle