[SPARC64]: Avoid membar instructions in delay slots.
In particular, avoid membar instructions in the delay
slot of a jmpl instruction.
UltraSPARC-I, II, IIi, and IIe have a bug, documented in
the UltraSPARC-IIi User's Manual, Appendix K, Erratum 51
The long and short of it is that if the IMU unit misses
on a branch or jmpl, and there is a store buffer synchronizing
membar in the delay slot, the chip can stop fetching instructions.
If interrupts are enabled or some other trap is enabled, the
chip will unwedge itself, but performance will suffer.
We already had a workaround for this bug in a few spots, but
it's better to have the entire tree sanitized for this rule.
Signed-off-by: David S. Miller <davem@davemloft.net>
diff --git a/include/asm-sparc64/rwsem.h b/include/asm-sparc64/rwsem.h
index bf2ae90..a1cc94f 100644
--- a/include/asm-sparc64/rwsem.h
+++ b/include/asm-sparc64/rwsem.h
@@ -55,8 +55,9 @@
"add %%g1, %1, %%g7\n\t"
"cas [%2], %%g1, %%g7\n\t"
"cmp %%g1, %%g7\n\t"
+ "membar #StoreLoad | #StoreStore\n\t"
"bne,pn %%icc, 1b\n\t"
- " membar #StoreLoad | #StoreStore\n\t"
+ " nop\n\t"
"mov %%g7, %0\n\t"
: "=&r" (tmp)
: "0" (tmp), "r" (sem)
diff --git a/include/asm-sparc64/spinlock.h b/include/asm-sparc64/spinlock.h
index db7581b..9cb93a5 100644
--- a/include/asm-sparc64/spinlock.h
+++ b/include/asm-sparc64/spinlock.h
@@ -52,12 +52,14 @@
__asm__ __volatile__(
"1: ldstub [%1], %0\n"
+" membar #StoreLoad | #StoreStore\n"
" brnz,pn %0, 2f\n"
-" membar #StoreLoad | #StoreStore\n"
+" nop\n"
" .subsection 2\n"
"2: ldub [%1], %0\n"
+" membar #LoadLoad\n"
" brnz,pt %0, 2b\n"
-" membar #LoadLoad\n"
+" nop\n"
" ba,a,pt %%xcc, 1b\n"
" .previous"
: "=&r" (tmp)
@@ -95,16 +97,18 @@
__asm__ __volatile__(
"1: ldstub [%2], %0\n"
-" brnz,pn %0, 2f\n"
" membar #StoreLoad | #StoreStore\n"
+" brnz,pn %0, 2f\n"
+" nop\n"
" .subsection 2\n"
"2: rdpr %%pil, %1\n"
" wrpr %3, %%pil\n"
"3: ldub [%2], %0\n"
-" brnz,pt %0, 3b\n"
" membar #LoadLoad\n"
+" brnz,pt %0, 3b\n"
+" nop\n"
" ba,pt %%xcc, 1b\n"
-" wrpr %1, %%pil\n"
+" wrpr %1, %%pil\n"
" .previous"
: "=&r" (tmp1), "=&r" (tmp2)
: "r"(lock), "r"(flags)
@@ -162,12 +166,14 @@
"4: add %0, 1, %1\n"
" cas [%2], %0, %1\n"
" cmp %0, %1\n"
+" membar #StoreLoad | #StoreStore\n"
" bne,pn %%icc, 1b\n"
-" membar #StoreLoad | #StoreStore\n"
+" nop\n"
" .subsection 2\n"
"2: ldsw [%2], %0\n"
+" membar #LoadLoad\n"
" brlz,pt %0, 2b\n"
-" membar #LoadLoad\n"
+" nop\n"
" ba,a,pt %%xcc, 4b\n"
" .previous"
: "=&r" (tmp1), "=&r" (tmp2)
@@ -204,12 +210,14 @@
"4: or %0, %3, %1\n"
" cas [%2], %0, %1\n"
" cmp %0, %1\n"
+" membar #StoreLoad | #StoreStore\n"
" bne,pn %%icc, 1b\n"
-" membar #StoreLoad | #StoreStore\n"
+" nop\n"
" .subsection 2\n"
"2: lduw [%2], %0\n"
+" membar #LoadLoad\n"
" brnz,pt %0, 2b\n"
-" membar #LoadLoad\n"
+" nop\n"
" ba,a,pt %%xcc, 4b\n"
" .previous"
: "=&r" (tmp1), "=&r" (tmp2)
@@ -240,8 +248,9 @@
" or %0, %4, %1\n"
" cas [%3], %0, %1\n"
" cmp %0, %1\n"
+" membar #StoreLoad | #StoreStore\n"
" bne,pn %%icc, 1b\n"
-" membar #StoreLoad | #StoreStore\n"
+" nop\n"
" mov 1, %2\n"
"2:"
: "=&r" (tmp1), "=&r" (tmp2), "=&r" (result)
diff --git a/include/asm-sparc64/spitfire.h b/include/asm-sparc64/spitfire.h
index 9d7613e..1aa9327 100644
--- a/include/asm-sparc64/spitfire.h
+++ b/include/asm-sparc64/spitfire.h
@@ -111,7 +111,6 @@
"membar #Sync"
: /* No outputs */
: "r" (tag), "r" (addr), "i" (ASI_DCACHE_TAG));
- __asm__ __volatile__ ("membar #Sync" : : : "memory");
}
/* The instruction cache lines are flushed with this, but note that