Yes, copying that file from a machine that doesn't have that LLVM bug would work around this issue.
I don't see an easy way for us to avoid using lli. We need to bootstrap our system info here. Perhaps we could run a stripped down version of emscripten to do this, not sure how easy that would be. - Alon On Sat, Oct 18, 2014 at 4:18 AM, Bruce Mitchener <[email protected]> wrote: > The info being constructed is for the EMSCRIPTEN target and not the build > host. The info is always the same unless emscripten changes a header or > compiler. > > Sent from my iPhone > > On Oct 18, 2014, at 6:03 PM, Anton Smirnov <[email protected]> wrote: > > Probably you're right. But on other host it might be another headers set > or struct sizes. Is there any opportunity to do it on host without using > lli? > > суббота, 18 октября 2014 г., 16:52:23 UTC+6 пользователь Bruce Mitchener > написал: >> >> You could copy the generated file from another host so that you don't >> need Emscripten on your target device to run lli ... >> >> - Bruce >> >> Sent from my iPhone >> >> On Oct 18, 2014, at 5:09 PM, Anton Smirnov <[email protected]> >> wrote: >> >> i've found emscripten uses `lli` tool (gen_struct_info.py): >> >> # NOTE: We can't generate an executable in the next step because it won't >> run on the current system without changing the target. >> # If we change the target, some type sizes will change resulting in wrong >> data. As a workaround, we will be generating bitcode and >> # run that with the LLVM interpreter. That way we can use the default >> target and still run the code. >> >> Can anybody please explain to me that hack? Can we avoid using lli tool >> (probably do the same in python script or in js)? >> The problem is that on some platforms (ios f.e.) lli can't work because >> of exception: >> >> "args": [ >> "lli", >> "/private/var/mobile/Applications/10BD667C-339F- >> 4A25-B6BF-D5404F079E5C/tmp/tmpOpO_KF.ll" >> >> raises exception in ExecutionEngine in : >> >> return runFunction(Fn, GVArgs).IntVal.getZExtValue(); >> >> exc_bad_instruction (code=exc_arm_undefined, subcode=0xf4cffce8). >> >> Same unresolved issue in llvm: >> >> http://llvm.org/bugs/show_bug.cgi?id=21012 >> >> Any thoughts are highly appreciated. >> >> >> -- >> You received this message because you are subscribed to the Google Groups >> "emscripten-discuss" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to [email protected]. >> For more options, visit https://groups.google.com/d/optout. >> >> -- > You received this message because you are subscribed to the Google Groups > "emscripten-discuss" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit https://groups.google.com/d/optout. > > -- > You received this message because you are subscribed to the Google Groups > "emscripten-discuss" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit https://groups.google.com/d/optout. > -- You received this message because you are subscribed to the Google Groups "emscripten-discuss" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
