diff options
| author | Yonghong Song <yhs@fb.com> | 2017-10-16 04:14:53 +0000 |
|---|---|---|
| committer | Yonghong Song <yhs@fb.com> | 2017-10-16 04:14:53 +0000 |
| commit | e53750e1e086f4e234f52041072e688abd79e22b (patch) | |
| tree | 97d2e42448831ef3325e94f03deff012614d1e64 /test | |
| parent | a7d4828a91f5e9f6a77369ace1e9f37804b8ffad (diff) | |
bpf: fix bug on silently truncating 64-bit immediate
We came across an llvm bug when compiling some testcases that 64-bit
immediates are silently truncated into 32-bit and then packed into
BPF_JMP | BPF_K encoding. This caused comparison with wrong value.
This bug looks to be introduced by r308080. The Select_Ri pattern is
supposed to be lowered into J*_Ri while the latter only support 32-bit
immediate encoding, therefore Select_Ri should have similar immediate
predicate check as what J*_Ri are doing.
Reported-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Jiong Wang <jiong.wang@netronome.com>
Reviewed-by: Yonghong Song <yhs@fb.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@315889 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test')
| -rw-r--r-- | test/CodeGen/BPF/select_ri.ll | 35 |
1 files changed, 35 insertions, 0 deletions
diff --git a/test/CodeGen/BPF/select_ri.ll b/test/CodeGen/BPF/select_ri.ll index b802b64b728..7b1f852ca79 100644 --- a/test/CodeGen/BPF/select_ri.ll +++ b/test/CodeGen/BPF/select_ri.ll @@ -25,3 +25,38 @@ entry: } attributes #0 = { norecurse nounwind readonly } + +; test immediate out of 32-bit range +; Source file: + +; unsigned long long +; load_word(void *buf, unsigned long long off) +; asm("llvm.bpf.load.word"); +; +; int +; foo(void *buf) +; { +; unsigned long long sum = 0; +; +; sum += load_word(buf, 100); +; sum += load_word(buf, 104); +; +; if (sum != 0x1ffffffffULL) +; return ~0U; +; +; return 0; +;} + +; Function Attrs: nounwind readonly +define i32 @foo(i8*) local_unnamed_addr #0 { + %2 = tail call i64 @llvm.bpf.load.word(i8* %0, i64 100) + %3 = tail call i64 @llvm.bpf.load.word(i8* %0, i64 104) + %4 = add i64 %3, %2 + %5 = icmp ne i64 %4, 8589934591 +; CHECK: r{{[0-9]+}} = 8589934591 ll + %6 = sext i1 %5 to i32 + ret i32 %6 +} + +; Function Attrs: nounwind readonly +declare i64 @llvm.bpf.load.word(i8*, i64) #1 |