compile error under mingw 4.8.1 about jemalloc HOT 4 CLOSED

• ffs() and ffsl() are defined for MinGW in include/msvc_compat/strings.h. I don't see any evidence of build system regression that would cause that header to be missed by #include <strings.h>. Is there something unusual about the environment you have specified when calling the configure script?
• Initialization of init_lock is taken care of in _init_init_lock(), which has the constructor attribute, and therefore gets run prior to malloc_init_hard(). What is the nature of the error you're seeing?
• All of the %z format specifiers are either %zu or %zd, so I'm not sure what %z/size_t mismatch you're referring to.

I don't have a Windows environment set up, and Microsoft makes it so difficult/expensive that I'm not likely to do so in time for the 3.5.0 release. If you need help developing fixes, you might try engaging others who use jemalloc with MinGW on the jemalloc-discuss mailing list.

from jemalloc.

No thanks, I fixed these errors and jemalloc is working fine. I just want to report these errors maybe someone want to fix it.
I don't want to commit my code, because I just hacked the code instead of fixing these errors. So my patches can ruin the code on other platforms. I don't know how to fix it properly.

• configure is not included include/msvc_compat/strings.h. I think this header is inlcuding only under VS.
• InitializeCriticalSection() or InitializeCriticalSectionAndSpinCount() must call before EnterCriticalSection() called, but in malloc_init_hard() is not fulfilled.
• you right, %zu or %zd has been changed to %u and %d.

You can close this issue there are not many people who want to use it under mingw.

from jemalloc.

Maybe this line in configure.ac should be executed for all MinGW builds:

CPPFLAGS="$CPPFLAGS -I${srcroot}/include/msvc_compat"

I'm going to leave this issue open until someone is able to evaluate what fixes are actually necessary to support both MSVC and gcc under MinGW. Thank you for the report.

from jemalloc.

These issues have all been fixed now, via the addition of jemalloc_ffs() and PRIzu/PRIzd/PRIzx.

from jemalloc.