runtime, reflect: use correctly aligned stack frame sizes on arm64

arm64 requires either no stack frame or a frame with a size that is 8 mod 16
(adding the saved LR will make it 16-aligned).

The cmd/internal/obj/arm64 has been silently aligning frames, but it led to
a terrible bug when the compiler and obj disagreed on the frame size,
and it's just generally confusing, so we're going to make misaligned frames
an error instead of something that is silently changed.

This CL prepares by updating assembly files.
Note that the changes in this CL are already being done silently by
cmd/internal/obj/arm64, so there is no semantic effect here,
just a clarity effect.

For #9880.

Change-Id: Ibd6928dc5fdcd896c2bacd0291bf26b364591e28
Reviewed-on: https://go-review.googlesource.com/12845
Reviewed-by: Austin Clements <austin@google.com>

diff --git a/src/reflect/asm_arm64.s b/src/reflect/asm_arm64.s
index a5a8b91c1ea56c6ae85834754b7a68bf7bd94017..bdd38438aec86f297fa50ad9f8c43c796efe9264 100644 (file)
--- a/src/reflect/asm_arm64.s
+++ b/src/reflect/asm_arm64.s
@@ -9,7 +9,7 @@
 // See the comment on the declaration of makeFuncStub in makefunc.go
 // for more details.
 // No arg size here, runtime pulls arg map out of the func value.
-TEXT ·makeFuncStub(SB),(NOSPLIT|WRAPPER),$16
+TEXT ·makeFuncStub(SB),(NOSPLIT|WRAPPER),$24
        NO_LOCAL_POINTERS
        MOVD    R26, 8(RSP)
        MOVD    $argframe+0(FP), R3
@@ -21,7 +21,7 @@ TEXT ·makeFuncStub(SB),(NOSPLIT|WRAPPER),$16
 // See the comment on the declaration of methodValueCall in makefunc.go
 // for more details.
 // No arg size here; runtime pulls arg map out of the func value.
-TEXT ·methodValueCall(SB),(NOSPLIT|WRAPPER),$16
+TEXT ·methodValueCall(SB),(NOSPLIT|WRAPPER),$24
        NO_LOCAL_POINTERS
        MOVD    R26, 8(RSP)
        MOVD    $argframe+0(FP), R3

diff --git a/src/runtime/asm_arm64.s b/src/runtime/asm_arm64.s
index 50094db1800344ed69b98054ff1b88d5ee789db8..9aff9c7e6c8c0d9aef97bed3a2bbf93e06df2093 100644 (file)
--- a/src/runtime/asm_arm64.s
+++ b/src/runtime/asm_arm64.s
@@ -424,33 +424,35 @@ end:                                              \
        BL      runtime·callwritebarrier(SB);  \
        RET
 
-CALLFN(·call16, 16)
-CALLFN(·call32, 32)
-CALLFN(·call64, 64)
-CALLFN(·call128, 128)
-CALLFN(·call256, 256)
-CALLFN(·call512, 512)
-CALLFN(·call1024, 1024)
-CALLFN(·call2048, 2048)
-CALLFN(·call4096, 4096)
-CALLFN(·call8192, 8192)
-CALLFN(·call16384, 16384)
-CALLFN(·call32768, 32768)
-CALLFN(·call65536, 65536)
-CALLFN(·call131072, 131072)
-CALLFN(·call262144, 262144)
-CALLFN(·call524288, 524288)
-CALLFN(·call1048576, 1048576)
-CALLFN(·call2097152, 2097152)
-CALLFN(·call4194304, 4194304)
-CALLFN(·call8388608, 8388608)
-CALLFN(·call16777216, 16777216)
-CALLFN(·call33554432, 33554432)
-CALLFN(·call67108864, 67108864)
-CALLFN(·call134217728, 134217728)
-CALLFN(·call268435456, 268435456)
-CALLFN(·call536870912, 536870912)
-CALLFN(·call1073741824, 1073741824)
+// These have 8 added to make the overall frame size a multiple of 16,
+// as required by the ABI. (There is another +8 for the saved LR.)
+CALLFN(·call16, 24 )
+CALLFN(·call32, 40 )
+CALLFN(·call64, 72 )
+CALLFN(·call128, 136 )
+CALLFN(·call256, 264 )
+CALLFN(·call512, 520 )
+CALLFN(·call1024, 1032 )
+CALLFN(·call2048, 2056 )
+CALLFN(·call4096, 4104 )
+CALLFN(·call8192, 8200 )
+CALLFN(·call16384, 16392 )
+CALLFN(·call32768, 32776 )
+CALLFN(·call65536, 65544 )
+CALLFN(·call131072, 131080 )
+CALLFN(·call262144, 262152 )
+CALLFN(·call524288, 524296 )
+CALLFN(·call1048576, 1048584 )
+CALLFN(·call2097152, 2097160 )
+CALLFN(·call4194304, 4194312 )
+CALLFN(·call8388608, 8388616 )
+CALLFN(·call16777216, 16777224 )
+CALLFN(·call33554432, 33554440 )
+CALLFN(·call67108864, 67108872 )
+CALLFN(·call134217728, 134217736 )
+CALLFN(·call268435456, 268435464 )
+CALLFN(·call536870912, 536870920 )
+CALLFN(·call1073741824, 1073741832 )
 
 // bool cas(uint32 *ptr, uint32 old, uint32 new)
 // Atomically:
@@ -613,7 +615,7 @@ TEXT runtime·cgocallback(SB),NOSPLIT,$24-24
 
 // cgocallback_gofunc(FuncVal*, void *frame, uintptr framesize)
 // See cgocall.go for more details.
-TEXT ·cgocallback_gofunc(SB),NOSPLIT,$16-24
+TEXT ·cgocallback_gofunc(SB),NOSPLIT,$24-24
        NO_LOCAL_POINTERS
 
        // Load g from thread-local storage.
@@ -721,7 +723,7 @@ droppedm:
 
 // Called from cgo wrappers, this function returns g->m->curg.stack.hi.
 // Must obey the gcc calling convention.
-TEXT _cgo_topofstack(SB),NOSPLIT,$16
+TEXT _cgo_topofstack(SB),NOSPLIT,$24
        // g (R28) and REGTMP (R27)  might be clobbered by load_g. They
        // are callee-save in the gcc calling convention, so save them.
        MOVD    R27, savedR27-8(SP)

diff --git a/src/runtime/sys_linux_arm64.s b/src/runtime/sys_linux_arm64.s
index 998484f121fa4e5f4c57c2380f43254eb0bada0a..7b58d67305bc782e0740b19f5708dcdc7e6c36d0 100644 (file)
--- a/src/runtime/sys_linux_arm64.s
+++ b/src/runtime/sys_linux_arm64.s
@@ -115,7 +115,7 @@ TEXT runtime·getrlimit(SB),NOSPLIT,$-8-20
        MOVW    R0, ret+16(FP)
        RET
 
-TEXT runtime·usleep(SB),NOSPLIT,$16-4
+TEXT runtime·usleep(SB),NOSPLIT,$24-4
        MOVWU   usec+0(FP), R3
        MOVD    R3, R5
        MOVW    $1000000, R4
@@ -180,7 +180,7 @@ TEXT runtime·mincore(SB),NOSPLIT,$-8-28
        RET
 
 // func now() (sec int64, nsec int32)
-TEXT time·now(SB),NOSPLIT,$16-12
+TEXT time·now(SB),NOSPLIT,$24-12
        MOVD    RSP, R0
        MOVD    $0, R1
        MOVD    $SYS_gettimeofday, R8
@@ -193,7 +193,7 @@ TEXT time·now(SB),NOSPLIT,$16-12
        MOVW    R5, nsec+8(FP)
        RET
 
-TEXT runtime·nanotime(SB),NOSPLIT,$16-8
+TEXT runtime·nanotime(SB),NOSPLIT,$24-8
        MOVW    $1, R0 // CLOCK_MONOTONIC
        MOVD    RSP, R1
        MOVD    $SYS_clock_gettime, R8