powerpc/book3e: Fix single step when using HW page tables

We patch the TLB miss exception vectors to point to alternate
functions when using HW page table on BookE.

However, we were patching in a new branch in the first instruction
of the exception handler instead of the second one, thus overriding
the nop that is in the first instruction.

This cause problems when single stepping as we rely on that nop for
the single step to stop properly within the exception vector range
rather than on the target of the branch.
Signed-off-by: default avatarBenjamin Herrenschmidt <benh@kernel.crashing.org>
parent 34d97e07
......@@ -191,6 +191,12 @@ exc_##n##_bad_stack: \
sth r1,PACA_TRAP_SAVE(r13); /* store trap */ \
b bad_stack_book3e; /* bad stack error */
/* WARNING: If you change the layout of this stub, make sure you chcek
* the debug exception handler which handles single stepping
* into exceptions from userspace, and the MM code in
* arch/powerpc/mm/tlb_nohash.c which patches the branch here
* and would need to be updated if that branch is moved
*/
#define EXCEPTION_STUB(loc, label) \
. = interrupt_base_book3e + loc; \
nop; /* To make debug interrupts happy */ \
......
......@@ -391,10 +391,15 @@ static void __early_init_mmu(int boot_cpu)
/* Check if HW loader is supported */
if ((tlb0cfg & TLBnCFG_IND) &&
(tlb0cfg & TLBnCFG_PT)) {
patch_branch(ibase + (0x1c0 / 4),
(unsigned long)&exc_data_tlb_miss_htw_book3e, 0);
patch_branch(ibase + (0x1e0 / 4),
(unsigned long)&exc_instruction_tlb_miss_htw_book3e, 0);
/* Our exceptions vectors start with a NOP and -then- a branch
* to deal with single stepping from userspace which stops on
* the second instruction. Thus we need to patch the second
* instruction of the exception, not the first one
*/
patch_branch(ibase + (0x1c0 / 4) + 1,
(unsigned long)&exc_data_tlb_miss_htw_book3e, 0);
patch_branch(ibase + (0x1e0 / 4) + 1,
(unsigned long)&exc_instruction_tlb_miss_htw_book3e, 0);
book3e_htw_enabled = 1;
}
pr_info("MMU: Book3E Page Tables %s\n",
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment