Skip to content
  1. Jul 14, 2015
  3. Jul 10, 2015
    • John David Anglin's avatar
      parisc: Fix some PTE/TLB race conditions and optimize __flush_tlb_range based on timing results · 01ab6057
      John David Anglin authored 
      

The increased use of pdtlb/pitlb instructions seemed to increase the
frequency of random segmentation faults building packages. Further, we
had a number of cases where TLB inserts would repeatedly fail and all
forward progress would stop. The Haskell ghc package caused a lot of
trouble in this area. The final indication of a race in pte handling was
this syslog entry on sibaris (C8000):

 swap_free: Unused swap offset entry 00000004
 BUG: Bad page map in process mysqld  pte:00000100 pmd:019bbec5
 addr:00000000ec464000 vm_flags:00100073 anon_vma:0000000221023828 mapping: (null) index:ec464
 CPU: 1 PID: 9176 Comm: mysqld Not tainted 4.0.0-2-parisc64-smp #1 Debian 4.0.5-1
 Backtrace:
  [<0000000040173eb0>] show_stack+0x20/0x38
  [<0000000040444424>] dump_stack+0x9c/0x110
  [<00000000402a0d38>] print_bad_pte+0x1a8/0x278
  [<00000000402a28b8>] unmap_single_vma+0x3d8/0x770
  [<00000000402a4090>] zap_page_range+0xf0/0x198
  [<00000000402ba2a4>] SyS_madvise+0x404/0x8c0

Note that the pte value is 0 except for the accessed bit 0x100. This bit
shouldn't be set without the present bit.

It should be noted that the madvise system call is probably a trigger for many
of the random segmentation faults.

In looking at the kernel code, I found the following problems:

1) The pte_clear define didn't take TLB lock when clearing a pte.
2) We didn't test pte present bit inside lock in exception support.
3) The pte and tlb locks needed to merged in order to ensure consistency
between page table and TLB. This also has the effect of serializing TLB
broadcasts on SMP systems.

The attached change implements the above and a few other tweaks to try
to improve performance. Based on the timing code, TLB purges are very
slow (e.g., ~ 209 cycles per page on rp3440). Thus, I think it
beneficial to test the split_tlb variable to avoid duplicate purges.
Probably, all PA 2.0 machines have combined TLBs.

I dropped using __flush_tlb_range in flush_tlb_mm as I realized all
applications and most threads have a stack size that is too large to
make this useful. I added some comments to this effect.

Since implementing 1 through 3, I haven't had any random segmentation
faults on mx3210 (rp3440) in about one week of building code and running
as a Debian buildd.

Signed-off-by: default avatarJohn David Anglin <dave.anglin@bell.net>
Cc: stable@vger.kernel.org # v3.18+
Signed-off-by: default avatarHelge Deller <deller@gmx.de>
      01ab6057
    • Mark Rutland's avatar
      arm64: entry32: remove pointless register assignment · ad2daa85
      Mark Rutland authored 
      

We currently set x27 in compat_sys_sigreturn_wrapper and
compat_sys_rt_sigreturn_wrapper, similarly to what we do with r8/why on
32-bit ARM, in an attempt to prevent sigreturns from being restarted.

However, on arm64 we have always used pt_regs::syscallno for syscall
restarting (for both native and compat tasks), and x27 is never
inspected again before being overwritten in kernel_exit.

This patch removes the pointless register assignments.

Signed-off-by: default avatarMark Rutland <mark.rutland@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: default avatarCatalin Marinas <catalin.marinas@arm.com>
      ad2daa85
    • Ralf Baechle's avatar
      MIPS: O32: Use compat_sys_getsockopt. · 51d53674
      Ralf Baechle authored 
      

We were using the native syscall and that results in subtle breakage.

This is the same issue as fixed in 077d0e65
(MIPS: N32: Use compat getsockopt syscall) but that commit did fix it only
for N32.

Signed-off-by: default avatarRalf Baechle <ralf@linux-mips.org>
Link: https://bugzilla.kernel.org/show_bug.cgi?id=100291
      51d53674
    • Paul Burton's avatar
      MIPS: c-r4k: Extend way_string array · 1e18ac7a
      Paul Burton authored 
      

The L2 cache in the I6400 core has 16 ways, so extend the way_string
array to take such caches into account.

[ralf@linux-mips.org: Other already supported CPUs are free to support
more than 8 ways of cache as well.]

Signed-off-by: default avatarPaul Burton <paul.burton@imgtec.com>
Signed-off-by: default avatarMarkos Chandras <markos.chandras@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/10640/


Signed-off-by: default avatarRalf Baechle <ralf@linux-mips.org>
      1e18ac7a
    • James Hogan's avatar
      MIPS: Pistachio: Support CDMM & Fast Debug Channel · 6b5e741e
      James Hogan authored 
      

Implement the mips_cdmm_phys_base() platform callback to provide a
default Common Device Memory Map (CDMM) physical base address for the
Pistachio SoC. This allows the CDMM in each VPE to be configured and
probed for devices, such as the Fast Debug Channel (FDC).

The physical address chosen is just below the default CPC address, which
appears to also be unallocated.

The FDC IRQ is also usable on Pistachio, and is routed through the GIC,
so implement the get_c0_fdc_int() platform callback using
gic_get_c0_fdc_int(), so the FDC driver doesn't have to fall back to
polling.

Signed-off-by: default avatarJames Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Andrew Bresticker <abrestic@chromium.org>
Cc: James Hartley <james.hartley@imgtec.com>
Cc: linux-mips@linux-mips.org
Reviewed-by: default avatarAndrew Bresticker <abrestic@chromium.org>
Patchwork: http://patchwork.linux-mips.org/patch/9749/


Signed-off-by: default avatarRalf Baechle <ralf@linux-mips.org>
      6b5e741e
    • James Hogan's avatar
      MIPS: Malta: Make GIC FDC IRQ workaround Malta specific · 6249ecbb
      James Hogan authored 
      

Wider testing reveals that the Fast Debug Channel (FDC) interrupt is
routed through the GIC just fine on Pistachio SoC, even though it
contains interAptiv cores. Clearly the FDC interrupt routing problems
previously observed on interAptiv and proAptiv cores are specific to the
Malta FPGA bitstreams.

Move the workaround for interAptiv and proAptiv out of
gic_get_c0_fdc_int() in the GIC irqchip driver into Malta's
get_c0_fdc_int() platform callback, to allow the Pistachio SoC to use
the FDC interrupt.

Signed-off-by: default avatarJames Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Andrew Bresticker <abrestic@chromium.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Jason Cooper <jason@lakedaemon.net>
Cc: linux-mips@linux-mips.org
Reviewed-by: default avatarAndrew Bresticker <abrestic@chromium.org>
Cc: James Hartley <james.hartley@imgtec.com>
Patchwork: http://patchwork.linux-mips.org/patch/9748/


Signed-off-by: default avatarRalf Baechle <ralf@linux-mips.org>
      6249ecbb
    • Markos Chandras's avatar
      MIPS: c-r4k: Fix cache flushing for MT cores · cccf34e9
      Markos Chandras authored 
      

MT_SMP is not the only SMP option for MT cores. The MT_SMP option
allows more than one VPE per core to appear as a secondary CPU in the
system. Because of how CM works, it propagates the address-based
cache ops to the secondary cores but not the index-based ones.
Because of that, the code does not use IPIs to flush the L1 caches on
secondary cores because the CM would have done that already. However,
the CM functionality is independent of the type of SMP kernel so even in
non-MT kernels, IPIs are not necessary. As a result of which, we change
the conditional to depend on the CM presence. Moreover, since VPEs on
the same core share the same L1 caches, there is no need to send an
IPI on all of them so we calculate a suitable cpumask with only one
VPE per core.

Signed-off-by: default avatarMarkos Chandras <markos.chandras@imgtec.com>
Cc: <stable@vger.kernel.org> # 3.15+
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/10654/


Signed-off-by: default avatarRalf Baechle <ralf@linux-mips.org>
      cccf34e9
  5. Jul 09, 2015
  7. Jul 08, 2015
  9. Jul 07, 2015