Skip to content

K
Kernel

openeuler / Kernel
1 年多 前同步成功

通知 8
Star 0
Fork 0
K
Kernel
切换分支/标签
  1. 16 3月, 2016 4 次提交
    • C
      powerpc: Fix unrecoverable SLB miss during restore_math() · 6e669f08
      由 Cyril Bur 提交于 
      Commit 70fe3d98 "powerpc: Restore FPU/VEC/VSX if previously used" introduces a
call to restore_math() late in the syscall return path, after MSR_RI has been
cleared. The MSR_RI flag is used to indicate whether the kernel can take
another exception or not. A cleared MSR_RI flag indicates that the kernel
cannot.

Unfortunately when a machine is under SLB pressure an SLB miss can occur
in restore_math() which (with MSR_RI cleared) leads to an unrecoverable
exception.

  Unrecoverable exception 4100 at c0000000000088d8
  cpu 0x0: Vector: 4100  at [c0000003fa473b20]
      pc: c0000000000088d8: .load_vr_state+0x70/0x110
      lr: c00000000000f710: .restore_math+0x130/0x188
      sp: c0000003fa473da0
     msr: 9000000002003030
    current = 0xc0000007f876f180
    paca    = 0xc00000000fff0000	 softe: 0	 irq_happened: 0x01
      pid   = 1944, comm = K08umountfs
  [link register   ] c00000000000f710 .restore_math+0x130/0x188
  [c0000003fa473da0] c0000003fa473e30 (unreliable)
  [c0000003fa473e30] c000000000007b6c system_call+0x84/0xfc

The clearing of MSR_RI is actually an optimisation to avoid multiple MSR
writes, what must be disabled are interrupts. See comment in entry_64.S:

  /*
   * For performance reasons we clear RI the same time that we
   * clear EE. We only need to clear RI just before we restore r13
   * below, but batching it with EE saves us one expensive mtmsrd call.
   * We have to be careful to restore RI if we branch anywhere from
   * here (eg syscall_exit_work).
   */

At the point of calling restore_math() r13 has not been restored, as such, the
quick fix of turning MSR_RI back on for the call to restore_math() will
eliminate the occurrence of an unrecoverable exception.

We'd like to do a better fix in future.

Fixes: 70fe3d98 ("powerpc: Restore FPU/VEC/VSX if previously used")
Signed-off-by: NCyril Bur <cyrilbur@gmail.com>
Signed-off-by: NMichael Ellerman <mpe@ellerman.id.au>
      6e669f08
    • C
      powerpc/8xx: Fix do_mtspr_cpu6() build on older compilers · 2e098dce
      由 Christophe Leroy 提交于 
      GCC < 4.9 is unable to build this, saying:

  arch/powerpc/mm/8xx_mmu.c:139:2: error: memory input 1 is not directly addressable

Change the one-element array into a simple variable to avoid this.

Fixes: 1458dd95 ("powerpc/8xx: Handle CPU6 ERRATA directly in mtspr() macro")
Signed-off-by: NChristophe Leroy <christophe.leroy@c-s.fr>
Cc: Scott Wood <oss@buserror.net>
Signed-off-by: NMichael Ellerman <mpe@ellerman.id.au>
      2e098dce
    • M
      powerpc/rcpm: Fix build break when SMP=n · b081251e
      由 Michael Ellerman 提交于 
      Add an include of asm/smp.h to fix a build break when SMP=n:

  arch/powerpc/sysdev/fsl_rcpm.c:32:2: error: implicit declaration of
  function 'get_hard_smp_processor_id'

Fixes: d17799f9 ("powerpc/rcpm: add RCPM driver")
Signed-off-by: NMichael Ellerman <mpe@ellerman.id.au>
      b081251e
    • S
      powerpc/book3e-64: Use hardcoded mttmr opcode · 7a25d912
      由 Scott Wood 提交于 
      This preserves the ability to build using older binutils (reportedly <=
2.22).

Fixes: 6becef7e ("powerpc/mpc85xx: Add CPU hotplug support for E6500")
Signed-off-by: NScott Wood <oss@buserror.net>
Cc: chenhui.zhao@freescale.com
Signed-off-by: NMichael Ellerman <mpe@ellerman.id.au>
      7a25d912
  3. 14 3月, 2016 1 次提交
  5. 12 3月, 2016 30 次提交
  7. 11 3月, 2016 1 次提交
  9. 10 3月, 2016 4 次提交
    • M
      powerpc/perf: Fix misleading comment in pmao_restore_workaround() · 58bffb5b
      由 Madhavan Srinivasan 提交于 
      The current comment in pmao_restore_workaround() regarding
hard_irq_disable() is wrong. It should say to hard *disable* interrupts
instead of *enable*. Fix it.
Signed-off-by: NMadhavan Srinivasan <maddy@linux.vnet.ibm.com>
Signed-off-by: NMichael Ellerman <mpe@ellerman.id.au>
      58bffb5b
    • S
      powerpc/perf/24x7: Eliminate domain suffix in event names · 8f69dc70
      由 Sukadev Bhattiprolu 提交于 
      The Physical Core events of the 24x7 PMU can be monitored across various
domains (physical core, vcpu home core, vcpu home node etc). For each of
these core events, we currently create multiple events in sysfs, one for
each domain the event can be monitored in. These events are distinguished
by their suffixes like __PHYS_CORE, __VCPU_HOME_CORE etc.

Rather than creating multiple such entries, we could let the user specify
make 'domain' index a required parameter and let the user specify a value
for it (like they currently specify the core index).

	$ cat /sys/bus/event_source/devices/hv_24x7/events/HPM_CCYC
	domain=?,offset=0x98,core=?,lpar=0x0

	$ perf stat -C 0 -e hv_24x7/HPM_CCYC,domain=2,core=1/ true

(the 'domain=?' and 'core=?' in sysfs tell perf tool to enforce them as
required parameters).

This simplifies the interface and allows users to identify events by the
name specified in the catalog (User can determine the domain index by
referring to '/sys/bus/event_source/devices/hv_24x7/interface/domains').

Eliminating the event suffix eliminates several functions and simplifies
code.

Note that Physical Chip events can only be monitored in the chip domain
so those events have the domain set to 1 (rather than =?) and users don't
need to specify the domain index for the Chip events.

	$ cat /sys/bus/event_source/devices/hv_24x7/events/PM_XLINK_CYCLES
	domain=1,offset=0x230,chip=?,lpar=0x0

	$ perf stat -C 0 -e hv_24x7/PM_XLINK_CYCLES,chip=1/ true
Signed-off-by: NSukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: NMichael Ellerman <mpe@ellerman.id.au>
      8f69dc70
    • S
      powerpc/perf/hv-24x7: Display domain indices in sysfs · d34171e8
      由 Sukadev Bhattiprolu 提交于 
      To help users determine domains, display the domain indices used by the
kernel in sysfs.

	$ cat /sys/bus/event_source/devices/hv_24x7/interface/domains
	1: Physical Chip
	2: Physical Core
	3: VCPU Home Core
	4: VCPU Home Chip
	5: VCPU Home Node
	6: VCPU Remote Node
Signed-off-by: NSukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: NMichael Ellerman <mpe@ellerman.id.au>
      d34171e8
    • S
      powerpc/perf/hv-24x7: Display change in counter values · 2b206ee6
      由 Sukadev Bhattiprolu 提交于 
      For 24x7 counters, perf displays the raw value of the 24x7 counter, which
is a monotonically increasing value.

	perf stat -C 0 -e \
		'hv_24x7/HPM_0THRD_NON_IDLE_CCYC__PHYS_CORE,core=1/' \
		sleep 1

 Performance counter stats for 'CPU(s) 0':

     9,105,403,170      hv_24x7/HPM_0THRD_NON_IDLE_CCYC__PHYS_CORE,core=1/

       0.000425751 seconds time elapsed

In the typical usage of 'perf stat' this counter value is not as useful
as the _change_ in the counter value over the duration of the application.

Have h_24x7_event_init() set the event's prev_count to the raw value of
the 24x7 counter at the time of initialization. When the application
terminates, hv_24x7_event_read() will compute the change in value and
report to the perf tool. Similarly, for the transaction interface, clear
the event count to 0 at the beginning of the transaction.

	perf stat -C 0 -e \
		'hv_24x7/HPM_0THRD_NON_IDLE_CCYC__PHYS_CORE,core=1/' \
		sleep 1

 Performance counter stats for 'CPU(s) 0':

           245,758      hv_24x7/HPM_0THRD_NON_IDLE_CCYC__PHYS_CORE,core=1/

       1.006366383 seconds time elapsed
Signed-off-by: NSukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: NMichael Ellerman <mpe@ellerman.id.au>
      2b206ee6