<div dir="ltr"><div dir="ltr"><br></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Sat, Aug 15, 2020 at 8:39 PM Martin Storsjö <<a href="mailto:martin@martin.st">martin@martin.st</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Hi,<br>
<br>
<br>
On Sat, 15 Aug 2020, Ivan Serdyuk wrote:<br>
<br>
>       Just as Shoaib said, libunwind only is useful in environments<br>
>       that use<br>
>       the Itanium C++ ABI - there's really no use for it in an MSVC<br>
>       context<br>
>       (either using MSVC or clang-cl to compile it).<br>
><br>
>       The particular linker error comes from the fact that there's<br>
>       functions<br>
>       implemented in assembly, that expect the function name to be<br>
>       mangled the<br>
>       itanium way, while the object files built by the compiler expect<br>
>       the<br>
>       symbols to use the MSVC C++ name mangling, so there's an<br>
>       undefined<br>
>       reference.<br>
> <br>
> <br>
> I see, thanks.<br>
> <br>
> Which options exist for MSVC, in sense an alternative to libunwind and<br>
> libbacktrace ?<br>
<br>
Well for libunwind, there's really no use for it in an MSVC setting. All <br>
the unwinding functionality is already built into the operating system, <br>
available via the Rtl*Unwind* functions.<br></blockquote><div><br></div><div>Martin,</div><div>you mean these functions</div><div><a href="https://docs.microsoft.com/en-us/windows/win32/api/winnt/nf-winnt-rtlunwind">https://docs.microsoft.com/en-us/windows/win32/api/winnt/nf-winnt-rtlunwind</a></div><div><a href="https://docs.microsoft.com/en-us/windows/win32/api/winnt/nf-winnt-rtlunwind2">https://docs.microsoft.com/en-us/windows/win32/api/winnt/nf-winnt-rtlunwind2</a></div><div><a href="https://docs.microsoft.com/en-us/windows/win32/api/winnt/nf-winnt-rtlunwindex">https://docs.microsoft.com/en-us/windows/win32/api/winnt/nf-winnt-rtlunwindex</a></div><div><a href="https://docs.microsoft.com/en-us/windows/win32/api/winnt/nf-winnt-rtlvirtualunwind">https://docs.microsoft.com/en-us/windows/win32/api/winnt/nf-winnt-rtlvirtualunwind</a></div><div>?<br> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
For libbacktrace, I guess the _Unwind_Backtrace function in libunwind <br>
might work and be useful, even if the rest of libunwind doesn't make sense <br>
in such a context.</blockquote><div><br></div><div>Ian,</div><div>please share your vision/propose some plan.<br></div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"> I haven't tested this function on windows though, and <br>
it'd require you to fix the linker error one way or another.<br></blockquote><div> </div><div>I see. <br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
Another alternative, requiring much less code, would be to just extract <br>
the SEH version of _Unwind_Backtrace function from libgcc [1]. I think <br>
that function is fairly independent from the rest of libgcc and from the <br>
rest of the surrounding file, and only calls Windows system APIs.<br></blockquote><div><br></div><div>Well, adding committers to the discussion.</div><div>Richard, Tristan, Bernd, Jon - please verify/elaborate.</div><div><br></div><div>Thanks in advance.</div><div>Ivan<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
// Martin<br>
<br>
[1] <a href="https://github.com/gcc-mirror/gcc/blob/master/libgcc/unwind-seh.c#L434" rel="noreferrer" target="_blank">https://github.com/gcc-mirror/gcc/blob/master/libgcc/unwind-seh.c#L434</a><br>
<br>
<br>
</blockquote></div></div>