diff options
author | Steven Rostedt <srostedt@redhat.com> | 2008-11-20 16:18:55 -0500 |
---|---|---|
committer | Ingo Molnar <mingo@elte.hu> | 2008-11-28 08:07:45 -0500 |
commit | c7b0d17366d6e04a11470fc8d85f9fbac02671b9 (patch) | |
tree | 1725700cb0d4e8b58aec37d64f350a093e64e3cd /arch | |
parent | 813b8520f5c240c71df55d14095a7b171de264ce (diff) |
powerpc: ftrace, do nothing in mcount call for dyn ftrace
Impact: quicken mcount calls that are not replaced by dyn ftrace
Dynamic ftrace no longer does on the fly recording of mcount locations.
The mcount locations are now found at compile time. The mcount
function no longer needs to store registers and call a stub function.
It can now just simply return.
Since there are some functions that do not get converted to a nop
(.init sections and other code that may disappear), this patch should
help speed up that code.
Also, the stub for mcount on PowerPC 32 can not be a simple branch
link register like it is on PowerPC 64. According to the ABI specification:
"The _mcount routine is required to restore the link register from
the stack so that the profiling code can be inserted transparently,
whether or not the profiled function saves the link register itself."
This means that we must restore the link register that was used
to make the call to mcount. The minimal mcount function for PPC32
ends up being:
mcount:
mflr r0
mtctr r0
lwz r0, 4(r1)
mtlr r0
bctr
Where we move the link register used to call mcount into the
ctr register, and then restore the link register from the stack.
Then we use the ctr register to jump back to the mcount caller.
The r0 register is free for us to use.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Diffstat (limited to 'arch')
-rw-r--r-- | arch/powerpc/kernel/entry_32.S | 40 | ||||
-rw-r--r-- | arch/powerpc/kernel/entry_64.S | 12 |
2 files changed, 9 insertions, 43 deletions
diff --git a/arch/powerpc/kernel/entry_32.S b/arch/powerpc/kernel/entry_32.S index 7ecc0d1855c3..6f7eb7e00c79 100644 --- a/arch/powerpc/kernel/entry_32.S +++ b/arch/powerpc/kernel/entry_32.S | |||
@@ -1162,39 +1162,17 @@ machine_check_in_rtas: | |||
1162 | #ifdef CONFIG_DYNAMIC_FTRACE | 1162 | #ifdef CONFIG_DYNAMIC_FTRACE |
1163 | _GLOBAL(mcount) | 1163 | _GLOBAL(mcount) |
1164 | _GLOBAL(_mcount) | 1164 | _GLOBAL(_mcount) |
1165 | stwu r1,-48(r1) | 1165 | /* |
1166 | stw r3, 12(r1) | 1166 | * It is required that _mcount on PPC32 must preserve the |
1167 | stw r4, 16(r1) | 1167 | * link register. But we have r0 to play with. We use r0 |
1168 | stw r5, 20(r1) | 1168 | * to push the return address back to the caller of mcount |
1169 | stw r6, 24(r1) | 1169 | * into the ctr register, restore the link register and |
1170 | mflr r3 | 1170 | * then jump back using the ctr register. |
1171 | stw r7, 28(r1) | 1171 | */ |
1172 | mfcr r5 | 1172 | mflr r0 |
1173 | stw r8, 32(r1) | ||
1174 | stw r9, 36(r1) | ||
1175 | stw r10,40(r1) | ||
1176 | stw r3, 44(r1) | ||
1177 | stw r5, 8(r1) | ||
1178 | subi r3, r3, MCOUNT_INSN_SIZE | ||
1179 | .globl mcount_call | ||
1180 | mcount_call: | ||
1181 | bl ftrace_stub | ||
1182 | nop | ||
1183 | lwz r6, 8(r1) | ||
1184 | lwz r0, 44(r1) | ||
1185 | lwz r3, 12(r1) | ||
1186 | mtctr r0 | 1173 | mtctr r0 |
1187 | lwz r4, 16(r1) | 1174 | lwz r0, 4(r1) |
1188 | mtcr r6 | ||
1189 | lwz r5, 20(r1) | ||
1190 | lwz r6, 24(r1) | ||
1191 | lwz r0, 52(r1) | ||
1192 | lwz r7, 28(r1) | ||
1193 | lwz r8, 32(r1) | ||
1194 | mtlr r0 | 1175 | mtlr r0 |
1195 | lwz r9, 36(r1) | ||
1196 | lwz r10,40(r1) | ||
1197 | addi r1, r1, 48 | ||
1198 | bctr | 1176 | bctr |
1199 | 1177 | ||
1200 | _GLOBAL(ftrace_caller) | 1178 | _GLOBAL(ftrace_caller) |
diff --git a/arch/powerpc/kernel/entry_64.S b/arch/powerpc/kernel/entry_64.S index e6d52845854f..b00982e0d1ee 100644 --- a/arch/powerpc/kernel/entry_64.S +++ b/arch/powerpc/kernel/entry_64.S | |||
@@ -888,18 +888,6 @@ _GLOBAL(enter_prom) | |||
888 | #ifdef CONFIG_DYNAMIC_FTRACE | 888 | #ifdef CONFIG_DYNAMIC_FTRACE |
889 | _GLOBAL(mcount) | 889 | _GLOBAL(mcount) |
890 | _GLOBAL(_mcount) | 890 | _GLOBAL(_mcount) |
891 | /* Taken from output of objdump from lib64/glibc */ | ||
892 | mflr r3 | ||
893 | stdu r1, -112(r1) | ||
894 | std r3, 128(r1) | ||
895 | subi r3, r3, MCOUNT_INSN_SIZE | ||
896 | .globl mcount_call | ||
897 | mcount_call: | ||
898 | bl ftrace_stub | ||
899 | nop | ||
900 | ld r0, 128(r1) | ||
901 | mtlr r0 | ||
902 | addi r1, r1, 112 | ||
903 | blr | 891 | blr |
904 | 892 | ||
905 | _GLOBAL(ftrace_caller) | 893 | _GLOBAL(ftrace_caller) |