Le 18/09/2014 17:15, Joakim Tjernlund a écrit :

Christophe Leroy [off-list ref] wrote on 2014/09/17 18:36:57:

quoted

Exception InstructionAccess does not exist on MPC8xx. No need to branch

there from somewhere else.

quoted

Handling can be done directly in InstructionTLBError Exception.

Signed-off-by: Christophe Leroy <redacted>

---
Changes in v2:
- None

Changes in v3:
- arch/powerpc/mm/fault.c uses the vector number, so make sure it

understand

quoted

the new ones.

  arch/powerpc/kernel/head_8xx.S | 17 +++++++----------
  arch/powerpc/mm/fault.c        |  1 +
  2 files changed, 8 insertions(+), 10 deletions(-)

diff --git a/arch/powerpc/kernel/head_8xx.S

b/arch/powerpc/kernel/head_8xx.S

quoted

index 3af6db1..fbe5d10 100644

--- a/arch/powerpc/kernel/head_8xx.S
+++ b/arch/powerpc/kernel/head_8xx.S

@@ -234,15 +234,9 @@ DataAccess:
     EXC_XFER_LITE(0x300, handle_page_fault)

  /* Instruction access exception.
- * This is "never generated" by the MPC8xx.  We jump to it for other
- * translation errors.
+ * This is "never generated" by the MPC8xx.
   */
-   . = 0x400
-InstructionAccess:
-   EXCEPTION_PROLOG
-   mr   r4,r12
-   mr   r5,r9
-   EXC_XFER_LITE(0x400, handle_page_fault)
+   EXCEPTION(0x400, InstructionAccess, unknown_exception, EXC_XFER_STD)

  /* External interrupt */
     EXCEPTION(0x500, HardwareInterrupt, do_IRQ, EXC_XFER_LITE)

@@ -382,7 +376,7 @@ InstructionTLBMiss:
  #endif
     mfspr   r10, SPRN_SPRG_SCRATCH2
     EXCEPTION_EPILOG_0
-   b   InstructionAccess
+   b   InstructionTLBError

     . = 0x1200
  DataStoreTLBMiss:

@@ -477,7 +471,10 @@ DataStoreTLBMiss:
   */
     . = 0x1300
  InstructionTLBError:
-   b   InstructionAccess
+   EXCEPTION_PROLOG
+   mr   r4,r12
+   mr   r5,r9
+   EXC_XFER_LITE(0x1300, handle_page_fault)

Still don't like that you change the vector number, every other ppc uses
the
standard number.

Can you not just lie here(EXC_XFER_LITE(0x400, handle_page_fault))?
Move the code to InstructionAccess too and let TLBError branch there.

My issue was that if I do EXC_XFER_LITE(0x400, handle_page_fault), I 
can't leave the
     EXCEPTION(0x400, InstructionAccess, unknown_exception, 
EXC_XFER_STD) at address .400
Otherwise, I get twice the same label.

What about the following patch then ? Would this be acceptable ?

@@ -234,15 +234,13 @@ DataAccess:
      EXC_XFER_LITE(0x300, handle_page_fault)

  /* Instruction access exception.
- * This is "never generated" by the MPC8xx.  We jump to it for other
- * translation errors.
+ * This is "never generated" by the MPC8xx.
   */
      . = 0x400
  InstructionAccess:
      EXCEPTION_PROLOG
-    mr    r4,r12
-    mr    r5,r9
-    EXC_XFER_LITE(0x400, handle_page_fault)
+    addi    r3,r1,STACK_FRAME_OVERHEAD
+    EXC_XFER_STD(0x401, unknown_exception)

  /* External interrupt */
      EXCEPTION(0x500, HardwareInterrupt, do_IRQ, EXC_XFER_LITE)

@@ -382,7 +380,7 @@ InstructionTLBMiss:
  #endif
      mfspr    r10, SPRN_SPRG_SCRATCH2
      EXCEPTION_EPILOG_0
-    b    InstructionAccess
+    b    InstructionTLBError

      . = 0x1200
  DataStoreTLBMiss:

@@ -477,7 +475,11 @@ DataStoreTLBMiss:
   */
      . = 0x1300
  InstructionTLBError:
-    b    InstructionAccess
+    EXCEPTION_PROLOG
+    mr    r4,r12
+    mr    r5,r9
+    /* 0x400 is InstructionAccess exception, needed by bad_page_fault() */
+    EXC_XFER_LITE(0x400, handle_page_fault)

  /* This is the data TLB error on the MPC8xx.  This could be due to
   * many reasons, including a dirty update to a pte.  We can catch that

If not I think you should keep the old/current way?

Same for DataTLB Miss/Error

quoted

  /* This is the data TLB error on the MPC8xx.  This could be due to
   * many reasons, including a dirty update to a pte.  We can catch that

diff --git a/arch/powerpc/mm/fault.c b/arch/powerpc/mm/fault.c
index 51ab9e7..4d63c96 100644
--- a/arch/powerpc/mm/fault.c
+++ b/arch/powerpc/mm/fault.c

@@ -526,6 +526,7 @@ void bad_page_fault(struct pt_regs *regs, unsigned

long address, int sig)

quoted

        break;
     case 0x400:
     case 0x480:
+   case 0x1300:
        printk(KERN_ALERT "Unable to handle kernel paging request for "
           "instruction fetch\n");
        break;
-- 
2.1.0