2010/2/8 Vadim Zhukov <[email protected]>: > Thank you for your attention. And sorry, but I think that your version is > wrong: in case of only one "X" you'll have "tries" set to 1 instead of > NUM_CHARS.
<sigh> Time to write some regress tests for mktemp obviously. Do you happen to have a program reliably demonstrates the original buffer underrun problem? (I couldn't get mktemp(1) to crash, but perhaps you have combination of arguments, options, and (perhaps) /etc/malloc.conf flags that does the trick every time?) Philip Guenther
