On Tue, Nov 21, 2023 at 04:48:29PM +0100, Mark Kettenis wrote:
> > Date: Tue, 21 Nov 2023 16:08:13 +0100
> > From: Tobias Heider <tobias.hei...@stusta.de>
> >
> > On Tue, Nov 21, 2023 at 12:23:08PM +0100, Mark Kettenis wrote:
> > > > Date: Tue, 21 Nov 2023 12:04:42 +0100
> > > > From: Tobias Heider <tobias.hei...@stusta.de>
> > > >
> > > > On Tue, Nov 21, 2023 at 11:56:18AM +0100, Mark Kettenis wrote:
> > > > > > Date: Tue, 21 Nov 2023 00:16:40 +0100
> > > > > > From: Tobias Heider <tobias.hei...@stusta.de>
> > > > > >
> > > > > > Diff below fixes make regress for libffi with arm64 BTI enabled.
> > > > > > The tricky part were two jump tables in ffi.c and sysV.S.
> > > > > >
> > > > > > ok?
> > > > >
> > > > > I think you missed the "computed goto" in ffi_closure_SYSV.
> > > > >
> > > > > Maybe we shouldn't add a "bti j" for the unused slots?
> > > >
> > > > Functionally yes, that should also work and makes more sense.
> > > > But then the brks would be a bit redundant. I wasn't sure if there
> > > > are tests or something that expects those breaks to be reachable.
> > >
> > > The brks still need to be there for those poor CPUs that don't
> > > implement BTI.
> > >
> >
> > Here is an updated diff where all reserved and unused entries
> > don't get bti j. make regress still passes.
>
> Looks like you missed my comment about ffi_closure_SYSV.
urgh yes, missed that. fixed in diff below
> Is the closures stuff not tested as part of make regress?
it seems like a lot of tests are skipped in our port. I'll see
if I can force it to run a few more.
# of expected passes 5822
# of unsupported tests 209
Index: Makefile
===================================================================
RCS file: /cvs/ports/devel/libffi/Makefile,v
retrieving revision 1.48
diff -u -p -r1.48 Makefile
--- Makefile 21 Sep 2023 09:49:57 -0000 1.48
+++ Makefile 21 Nov 2023 16:07:43 -0000
@@ -1,6 +1,7 @@
COMMENT= Foreign Function Interface
V= 3.4.4
+REVISION= 0
DISTNAME= libffi-$V
SHARED_LIBS += ffi 2.0 # 9.2
CATEGORIES= devel
Index: patches/patch-src_aarch64_ffi_c
===================================================================
RCS file: patches/patch-src_aarch64_ffi_c
diff -N patches/patch-src_aarch64_ffi_c
--- /dev/null 1 Jan 1970 00:00:00 -0000
+++ patches/patch-src_aarch64_ffi_c 21 Nov 2023 16:07:43 -0000
@@ -0,0 +1,76 @@
+Index: src/aarch64/ffi.c
+--- src/aarch64/ffi.c.orig
++++ src/aarch64/ffi.c
+@@ -390,47 +390,59 @@ extend_hfa_type (void *dest, void *src, int h)
+ "adr %0, 0f\n"
+ " add %0, %0, %1\n"
+ " br %0\n"
+-"0: ldp s16, s17, [%3]\n" /* S4 */
++"0: bti j\n" /* S4 */
++" ldp s16, s17, [%3]\n"
+ " ldp s18, s19, [%3, #8]\n"
+ " b 4f\n"
+-" ldp s16, s17, [%3]\n" /* S3 */
++" bti j\n" /* S3 */
++" ldp s16, s17, [%3]\n"
+ " ldr s18, [%3, #8]\n"
+ " b 3f\n"
+-" ldp s16, s17, [%3]\n" /* S2 */
++" bti j\n" /* S2 */
++" ldp s16, s17, [%3]\n"
+ " b 2f\n"
+ " nop\n"
+-" ldr s16, [%3]\n" /* S1 */
++" bti j\n" /* S1 */
++" ldr s16, [%3]\n"
+ " b 1f\n"
+ " nop\n"
+-" ldp d16, d17, [%3]\n" /* D4 */
++" bti j\n" /* D4 */
++" ldp d16, d17, [%3]\n"
+ " ldp d18, d19, [%3, #16]\n"
+ " b 4f\n"
+-" ldp d16, d17, [%3]\n" /* D3 */
++" bti j\n" /* D3 */
++" ldp d16, d17, [%3]\n"
+ " ldr d18, [%3, #16]\n"
+ " b 3f\n"
+-" ldp d16, d17, [%3]\n" /* D2 */
++" bti j\n" /* D2 */
++" ldp d16, d17, [%3]\n"
+ " b 2f\n"
+ " nop\n"
+-" ldr d16, [%3]\n" /* D1 */
++" bti j\n" /* D1 */
++" ldr d16, [%3]\n"
+ " b 1f\n"
+ " nop\n"
+-" ldp q16, q17, [%3]\n" /* Q4 */
++" bti j\n" /* Q4 */
++" ldp q16, q17, [%3]\n"
+ " ldp q18, q19, [%3, #32]\n"
+ " b 4f\n"
+-" ldp q16, q17, [%3]\n" /* Q3 */
++" bti j\n" /* Q3 */
++" ldp q16, q17, [%3]\n"
+ " ldr q18, [%3, #32]\n"
+ " b 3f\n"
+-" ldp q16, q17, [%3]\n" /* Q2 */
++" bti j\n" /* Q2 */
++" ldp q16, q17, [%3]\n"
+ " b 2f\n"
+ " nop\n"
+-" ldr q16, [%3]\n" /* Q1 */
++" bti j\n" /* Q1 */
++" ldr q16, [%3]\n"
+ " b 1f\n"
+ "4: str q19, [%2, #48]\n"
+ "3: str q18, [%2, #32]\n"
+ "2: str q17, [%2, #16]\n"
+ "1: str q16, [%2]"
+ : "=&r"(x0)
+- : "r"(f * 12), "r"(dest), "r"(src)
++ : "r"(f * 16), "r"(dest), "r"(src)
+ : "memory", "v16", "v17", "v18", "v19");
+ }
+ #endif
Index: patches/patch-src_aarch64_sysv_S
===================================================================
RCS file: patches/patch-src_aarch64_sysv_S
diff -N patches/patch-src_aarch64_sysv_S
--- /dev/null 1 Jan 1970 00:00:00 -0000
+++ patches/patch-src_aarch64_sysv_S 21 Nov 2023 16:07:43 -0000
@@ -0,0 +1,369 @@
+Index: src/aarch64/sysv.S
+--- src/aarch64/sysv.S.orig
++++ src/aarch64/sysv.S
+@@ -78,6 +78,7 @@ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+ cfi_startproc
+ CNAME(ffi_call_SYSV):
++ bti c
+ /* Sign the lr with x1 since that is where it will be stored */
+ SIGN_LR_WITH_REG(x1)
+
+@@ -138,78 +139,142 @@ CNAME(ffi_call_SYSV):
+ /* Save the return value as directed. */
+ adr x5, 0f
+ and w4, w4, #AARCH64_RET_MASK
+- add x5, x5, x4, lsl #3
++ add x5, x5, x4, lsl #4
+ br x5
+
+- /* Note that each table entry is 2 insns, and thus 8 bytes.
++ /* Note that each table entry is 4 insns, and thus 16 bytes.
+ For integer data, note that we're storing into ffi_arg
+ and therefore we want to extend to 64 bits; these types
+ have two consecutive entries allocated for them. */
+ .align 4
+-0: b 99f /* VOID */
++0: bti j /* VOID */
++ b 99f
++ nop
+ nop
+-1: str x0, [x3] /* INT64 */
++1: bti j /* INT64 */
++ str x0, [x3]
+ b 99f
+-2: stp x0, x1, [x3] /* INT128 */
++ nop
++2: bti j /* INT128 */
++ stp x0, x1, [x3]
+ b 99f
++ nop
+ 3: brk #1000 /* UNUSED */
+ b 99f
++ nop
++ nop
+ 4: brk #1000 /* UNUSED */
+ b 99f
++ nop
++ nop
+ 5: brk #1000 /* UNUSED */
+ b 99f
++ nop
++ nop
+ 6: brk #1000 /* UNUSED */
+ b 99f
++ nop
++ nop
+ 7: brk #1000 /* UNUSED */
+ b 99f
+-8: st4 { v0.s, v1.s, v2.s, v3.s }[0], [x3] /* S4 */
++ nop
++ nop
++8: bti j /* S4 */
++ st4 { v0.s, v1.s, v2.s, v3.s }[0], [x3]
+ b 99f
+-9: st3 { v0.s, v1.s, v2.s }[0], [x3] /* S3 */
++ nop
++9: bti j /* S3 */
++ st3 { v0.s, v1.s, v2.s }[0], [x3]
+ b 99f
+-10: stp s0, s1, [x3] /* S2 */
++ nop
++10: bti j /* S2 */
++ stp s0, s1, [x3]
+ b 99f
+-11: str s0, [x3] /* S1 */
++ nop
++11: bti j
++ str s0, [x3] /* S1 */
+ b 99f
+-12: st4 { v0.d, v1.d, v2.d, v3.d }[0], [x3] /* D4 */
++ nop
++12: bti j /* D4 */
++ st4 { v0.d, v1.d, v2.d, v3.d }[0], [x3]
+ b 99f
+-13: st3 { v0.d, v1.d, v2.d }[0], [x3] /* D3 */
++ nop
++13: bti j /* D3 */
++ st3 { v0.d, v1.d, v2.d }[0], [x3]
+ b 99f
+-14: stp d0, d1, [x3] /* D2 */
++ nop
++14: bti j /* D2 */
++ stp d0, d1, [x3]
+ b 99f
+-15: str d0, [x3] /* D1 */
++ nop
++15: bti j /* D1 */
++ str d0, [x3]
+ b 99f
+-16: str q3, [x3, #48] /* Q4 */
+ nop
+-17: str q2, [x3, #32] /* Q3 */
++16: bti j /* Q4 */
++ str q3, [x3, #48]
+ nop
+-18: stp q0, q1, [x3] /* Q2 */
++ nop
++17: bti j /* Q3 */
++ str q2, [x3, #32]
++ nop
++ nop
++18: bti j /* Q2 */
++ stp q0, q1, [x3]
+ b 99f
+-19: str q0, [x3] /* Q1 */
++ nop
++19: bti j /* Q1 */
++ str q0, [x3]
+ b 99f
+-20: uxtb w0, w0 /* UINT8 */
++ nop
++20: bti j /* UINT8 */
++ uxtb w0, w0
+ str x0, [x3]
++ nop
+ 21: b 99f /* reserved */
+ nop
+-22: uxth w0, w0 /* UINT16 */
++ nop
++ nop
++22: bti j /* UINT16 */
++ uxth w0, w0
+ str x0, [x3]
++ nop
+ 23: b 99f /* reserved */
+ nop
+-24: mov w0, w0 /* UINT32 */
++ nop
++ nop
++24: bti j /* UINT32 */
++ mov w0, w0
+ str x0, [x3]
++ nop
+ 25: b 99f /* reserved */
+ nop
+-26: sxtb x0, w0 /* SINT8 */
++ nop
++ nop
++26: bti j /* SINT8 */
++ sxtb x0, w0
+ str x0, [x3]
++ nop
+ 27: b 99f /* reserved */
+ nop
+-28: sxth x0, w0 /* SINT16 */
++ nop
++ nop
++28: bti j /* SINT16 */
++ sxth x0, w0
+ str x0, [x3]
++ nop
+ 29: b 99f /* reserved */
+ nop
+-30: sxtw x0, w0 /* SINT32 */
++ nop
++ nop
++30: bti j /* SINT32 */
++ sxtw x0, w0
+ str x0, [x3]
++ nop
+ 31: b 99f /* reserved */
+ nop
++ nop
++ nop
+
+ /* Return now that result has been populated. */
+ 99:
+@@ -246,6 +311,7 @@ CNAME(ffi_call_SYSV):
+ .align 4
+ CNAME(ffi_closure_SYSV_V):
+ cfi_startproc
++ bti c
+ SIGN_LR
+ stp x29, x30, [sp, #-ffi_closure_SYSV_FS]!
+ cfi_adjust_cfa_offset (ffi_closure_SYSV_FS)
+@@ -270,6 +336,7 @@ CNAME(ffi_closure_SYSV_V):
+ .align 4
+ cfi_startproc
+ CNAME(ffi_closure_SYSV):
++ bti c
+ SIGN_LR
+ stp x29, x30, [sp, #-ffi_closure_SYSV_FS]!
+ cfi_adjust_cfa_offset (ffi_closure_SYSV_FS)
+@@ -299,74 +366,136 @@ CNAME(ffi_closure_SYSV):
+ /* Load the return value as directed. */
+ adr x1, 0f
+ and w0, w0, #AARCH64_RET_MASK
+- add x1, x1, x0, lsl #3
++ add x1, x1, x0, lsl #4
+ add x3, sp, #16+CALL_CONTEXT_SIZE
+ br x1
+
+- /* Note that each table entry is 2 insns, and thus 8 bytes. */
++ /* Note that each table entry is 4 insns, and thus 16 bytes. */
+ .align 4
+-0: b 99f /* VOID */
++0: bti j /* VOID */
++ b 99f
+ nop
+-1: ldr x0, [x3] /* INT64 */
++ nop
++1: bti j /* INT64 */
++ ldr x0, [x3]
+ b 99f
+-2: ldp x0, x1, [x3] /* INT128 */
++ nop
++2: bti j /* INT128 */
++ ldp x0, x1, [x3]
+ b 99f
++ nop
+ 3: brk #1000 /* UNUSED */
+ nop
++ nop
++ nop
+ 4: brk #1000 /* UNUSED */
+ nop
++ nop
++ nop
+ 5: brk #1000 /* UNUSED */
+ nop
++ nop
++ nop
+ 6: brk #1000 /* UNUSED */
+ nop
++ nop
++ nop
+ 7: brk #1000 /* UNUSED */
+ nop
+-8: ldr s3, [x3, #12] /* S4 */
+ nop
+-9: ldr s2, [x3, #8] /* S3 */
+ nop
+-10: ldp s0, s1, [x3] /* S2 */
++8: bti j /* S4 */
++ ldr s3, [x3, #12]
++ nop
++ nop
++9: bti j /* S3 */
++ ldr s2, [x3, #8]
++ nop
++ nop
++10: bti j /* S2 */
++ ldp s0, s1, [x3]
+ b 99f
+-11: ldr s0, [x3] /* S1 */
++ nop
++11: bti j /* S1 */
++ ldr s0, [x3]
+ b 99f
+-12: ldr d3, [x3, #24] /* D4 */
+ nop
+-13: ldr d2, [x3, #16] /* D3 */
++12: bti j /* D4 */
++ ldr d3, [x3, #24]
+ nop
+-14: ldp d0, d1, [x3] /* D2 */
++ nop
++13: bti j /* D3 */
++ ldr d2, [x3, #16]
++ nop
++ nop
++14: bti j /* D2 */
++ ldp d0, d1, [x3]
+ b 99f
+-15: ldr d0, [x3] /* D1 */
++ nop
++15: bti j /* D1 */
++ ldr d0, [x3]
+ b 99f
+-16: ldr q3, [x3, #48] /* Q4 */
+ nop
+-17: ldr q2, [x3, #32] /* Q3 */
++16: bti j /* Q4 */
++ ldr q3, [x3, #48]
+ nop
+-18: ldp q0, q1, [x3] /* Q2 */
++ nop
++17: bti j /* Q3 */
++ ldr q2, [x3, #32]
++ nop
++ nop
++18: bti j /* Q2 */
++ ldp q0, q1, [x3]
+ b 99f
+-19: ldr q0, [x3] /* Q1 */
++ nop
++19: bti j /* Q1 */
++ ldr q0, [x3]
+ b 99f
+-20: ldrb w0, [x3, #BE(7)] /* UINT8 */
++ nop
++20: bti j /* UINT8 */
++ ldrb w0, [x3, #BE(7)]
+ b 99f
++ nop
+ 21: brk #1000 /* reserved */
+ nop
+-22: ldrh w0, [x3, #BE(6)] /* UINT16 */
++ nop
++ nop
++22: bti j /* UINT16 */
++ ldrh w0, [x3, #BE(6)]
+ b 99f
++ nop
+ 23: brk #1000 /* reserved */
+ nop
+-24: ldr w0, [x3, #BE(4)] /* UINT32 */
++ nop
++ nop
++24: bti j /* UINT32 */
++ ldr w0, [x3, #BE(4)]
+ b 99f
++ nop
+ 25: brk #1000 /* reserved */
+ nop
+-26: ldrsb x0, [x3, #BE(7)] /* SINT8 */
++ nop
++ nop
++26: bti j /* SINT8 */
++ ldrsb x0, [x3, #BE(7)]
+ b 99f
++ nop
+ 27: brk #1000 /* reserved */
+ nop
+-28: ldrsh x0, [x3, #BE(6)] /* SINT16 */
++ nop
++ nop
++28: bti j /* SINT16 */
++ ldrsh x0, [x3, #BE(6)]
+ b 99f
++ nop
+ 29: brk #1000 /* reserved */
+ nop
+-30: ldrsw x0, [x3, #BE(4)] /* SINT32 */
+ nop
++ nop
++30: bti j /* SINT32 */
++ ldrsw x0, [x3, #BE(4)]
++ nop
++ nop
+ 31: /* reserved */
+ 99: ldp x29, x30, [sp], #ffi_closure_SYSV_FS
+ cfi_adjust_cfa_offset (-ffi_closure_SYSV_FS)
+@@ -479,6 +608,7 @@ CNAME(ffi_closure_trampoline_table_page):
+ .align 4
+ CNAME(ffi_go_closure_SYSV_V):
+ cfi_startproc
++ bti c
+ stp x29, x30, [sp, #-ffi_closure_SYSV_FS]!
+ cfi_adjust_cfa_offset (ffi_closure_SYSV_FS)
+ cfi_rel_offset (x29, 0)
+@@ -502,6 +632,7 @@ CNAME(ffi_go_closure_SYSV_V):
+ .align 4
+ cfi_startproc
+ CNAME(ffi_go_closure_SYSV):
++ bti c
+ stp x29, x30, [sp, #-ffi_closure_SYSV_FS]!
+ cfi_adjust_cfa_offset (ffi_closure_SYSV_FS)
+ cfi_rel_offset (x29, 0)
