[llvm-dev] Re：[compiler-rt] Why interceptor_via_fun doesn't work for *alloc functions? (original) (raw)

chuanqi.xcq via llvm-dev llvm-dev at lists.llvm.org
Wed Jul 22 18:40:10 PDT 2020 

Hi, all

Yesterday I find that interceptor_via_fun doesn't work for aligned_alloc when I try to suppress a error from the check for aligned_malloc. And after I try to look into the implementation of AddressSanitizer, I find that it doesn't implement the suppression mechanism for *alloc functions. From the codes I see, only the str* functions could be suppressed by the interceptor_via_fun way. I want to ask why compiler-rt design in this way. If this is a porgram, I want to try to fix this.

Thanks,
Chuanqi

发件人：llvm-dev at lists.llvm.org <llvm-dev at lists.llvm.org> 发送时间：2020年7月22日(星期三) 15:59 收件人：llvm-dev at lists.llvm.org <llvm-dev at lists.llvm.org> 主　题：[llvm-dev] How to suppress error reports from the external libraries in compiler-rt?

Hi, every one.

When I change the compiler of a project from gcc to clang, I met a problem about aligned_alloc. aligned_alloc accepts two arguments, one for alignment and another for size. The standard document says that the size must be integral multiple of alignment. But the gcc allows more relaxing restriction.

Now I met a external library which depend on this behavior, which would fail when the ASanityChecker checks it. And I can not modify the codes in that library to generate a new library. So I want to suppres the asanity check for aligned_alloc. I know here is a document (https://clang.llvm.org/docs/AddressSanitizer.html#suppressing-reports-in-external-libraries). However, it doesn't work when I try to write:

interceptor_via_fun:aligned_alloc

I also tried to write the function name who call aligned_alooc. But it doesn't work, too.

I know I need to ask to something like compiler-rt dev mailing list. But I didn't find it. So I ask it here.

Thanks, Chuanqi

-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20200723/cbca769c/attachment.html>

More information about the llvm-dev mailing list