https://bugs.kde.org/show_bug.cgi?id=384732
--- Comment #10 from Philippe Waroquiers <philippe.waroqui...@skynet.be> --- (In reply to Daniel from comment #9) > Do you have any idea if this will/can be fixed? Nobody seems to work on this for the moment, so time to fix is unknown. It looks like the solaris specific code should/could be made more general, so as to work on linux also. As I understand, the solaris code is intercepting a subset of the (solaris) libc functions in the child, so as to replace the direct assignment of the child status code in the parent address space by a procedure that gives back the child code to the parent via a pipe. I am wondering if it would not be simpler to just replace fully the posix_spawn in vg_preload, as the solaris solution seems quite dependent on internal implementation of libc (e.g. the replacement of get_error and set_error). > Or do I have any possibility to work around this issue in my code? What you might do is to write a replacement procedure for posix_spawn/posis_spawnp, using the technique/code of glibc 2.24. See http://www.valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.wrapping (of course, do not call the original posix_spawn, just implement it fully in the wrapping functionà. Basically, this suggestion consists in implementing the alternative solution discussed above, but outside of valgrind. So, you might maybe envisage to just write it inside vg_prelaoded.c and provide a patch, if ever what I suggest makes sense :) -- You are receiving this mail because: You are watching all bug changes.
