|
From: <dan...@ya...> - 2001-07-03 21:15:39
|
Hello all, The "garbage" left in exe files by the un-patched binutils on W95 is more than just inelegant, it is potentially a security problem. IMHO, a patched version should be made available quickly. I think keeping the fseek-fixup local to binutils is the safest option for the time being. There are some thread-safety concerns and also I haven't done any testing with other stream output functions (eg [f]puts and [f]putc) following fseek on W95. Is the fseek-fwrite sequence the only problem? In the event, I have rebuilt binutils with Mumits patch applied and would like to release. Also, I put in Eric Kohl's fastcall contrib into binutils. I have used recent (010702) binutils snap. Should I make available as test release? Now, Eric's fastcall also needs GCC support (changes in config/i386 files). Should GCC-2.95.3 (fastcall-test) be made available. I have done the bootstrap, and have tested with my code to check if it introduced any new bugs and nothing yet. I am just starting to play with the fastcall attribute now in test code. But I can only test on one machine. Hence my suggestion to make test release of this as well as binutils. If I release patched 2.95.3, what is the consensus on our old friend DIR_SEPARATOR. Which way shall she lean? Or should she wobble? I will be going to the snow with kids next week, so I can do before 8 July or after 16 July. Danny _____________________________________________________________________________ http://messenger.yahoo.com.au - Yahoo! Messenger - Voice chat, mail alerts, stock quotes and favourite news and lots more! |