On Sun, Oct 25, 2009 at 04:24:57PM -0400, Charles Wilson wrote: >Charles Wilson wrote: > >> Tested on mingw32 and cygwin. > >It occurred to me that I hadn't explicitly tested the error path with >this new version, and sure enough, there's a slight problem. Here's the >fix (relative to the previous patch): > >--- pseudo-reloc.c.save 2009-10-25 15:45:43.595012200 -0400 >+++ pseudo-reloc.c 2009-10-25 16:10:08.236212200 -0400 >@@ -94,7 +94,7 @@ > wchar_t module[MAX_PATH]; > char * posix_module = NULL; > static const char * UNKNOWN_MODULE = "<unknown module>: "; >- static const char * CYGWIN_FAILURE_MSG = "Cygwin runtime failure: "; >+ static const char CYGWIN_FAILURE_MSG[] = "Cygwin runtime failure: "; > static const size_t CYGWIN_FAILURE_MSG_LEN = sizeof >(CYGWIN_FAILURE_MSG) - 1; > DWORD len; > DWORD done; >@@ -133,6 +133,8 @@ > sizeof(UNKNOWN_MODULE), &done, NULL); > WriteFile (errh, (PCVOID)buf, len, &done, NULL); > } >+ WriteFile (errh, (PCVOID)"\n", 1, &done, NULL); >+ > cygwin_internal (CW_EXIT_PROCESS, > STATUS_ILLEGAL_DLL_PSEUDO_RELOCATION, > 1);
I didn't go through this + previous patch in exhaustive detail but if you've tested it then I think it's fine to check in. cgf
