Hi all,
The following tests fail because my username (csabaraduly) contains "bar" :
Hi all,
The following tests fail because my username (csabaraduly) contains "bar" :
True, it could probably skip the filename printed at the beginning of readobj with something like this:
#REGEX1: Symbols [#REGEX1-NOT: foobaz
#REGEX1-NOT: bar
#REGEX1-NOT: rebar
Mind contributing a patch?
For ignore-undefined-symbols.s, the simplest fix ought to be to have the llvm-mc RUN line take the source from :
becomes
But in this case, llvm-symbolizer still prints the file as $CWD/ which seems like its own separate bug.
–paulr
You can try this patch:
$ git diff
diff --git a/llvm/test/tools/llvm-objcopy/ELF/regex.test b/llvm/test/tools/llvm-objcopy/ELF/regex.test
index 6fa590b5dc5…04438143a6a 100644
— a/llvm/test/tools/llvm-objcopy/ELF/regex.test
+++ b/llvm/test/tools/llvm-objcopy/ELF/regex.test
@@ -47,14 +47,14 @@ Symbols:
Value: 0x1010
Size: 8
-#REGEX1-NOT: foobaz
-#REGEX1-NOT: bar
-#REGEX1-NOT: rebar
+#REGEX1-NOT: {{Name: *foobaz}}
+#REGEX1-NOT: {{Name: *bar}}
+#REGEX1-NOT: {{Name: *rebar}}
-#REGEX2-NOT: foobaz
-#REGEX2: bar
-#REGEX2-NOT: rebar
+#REGEX2-NOT: {{Name: *foobaz}}
+#REGEX2: {{Name: *bar}}
+#REGEX2-NOT: {{Name: *rebar}}
-#REGEX3: foobaz
-#REGEX3: bar
-#REGEX3: rebar
+#REGEX3: {{Name: *foobaz}}
+#REGEX3: {{Name: *bar}}
+#REGEX3: {{Name: *rebar}}
btw, if that patch does work for you, perhaps you or someone else could land it. I’m packing right now and will be on the road for a few days, or else I’d do it myself.
hth…
don
I think any test update should be to change the patterns to include the start and end of a symbol name, if it is the symbol name, or some other kind of delimiter, depending on the context. The same goes for any positive test cases, to avoid any spurious passes.
Example:
could become:
James