Date: Mon, 16 Oct 1995 16:29:01 +0200 From: "Andrew V. Stesin" <stesin@elvisti.kiev.ua> To: hackers@freebsd.org Subject: From BSDi mailing list on gcc-2.6.3 bugs: (fwd) A brief note about cc and gcc Message-ID: <199510161429.QAA26398@office.elvisti.kiev.ua>
next in thread | raw e-mail | index | archive | help
X-Inj-Env-From: ktts!devnull%havoc.techno.ru@crocodil.monolit.kiev.ua Sun Oct 15 11:17:29 1995 To: bsdi-users@bsdi.BSDI.COM Subject: A brief note about cc and gcc From: Chris Torek <torek@BSDI.COM> Date: Sat, 14 Oct 1995 19:49:29 -0600 In 2.0, `cc' gets you gcc 1.42 and `gcc' or `gcc2' gets you 2.6.3. Neither one of them is quite the same as the distribution version; in particular, I have installed at least a half dozen bug fixes in the 2.6.3 that we ship (including one that threw in a new bug, so that `gcc -pipe' does not always wait for the assembler to finish). I have not saved all of my test cases, but some sample bugs fixed: incorrect comparision of unsigned shorts incorrect loop jump optimization sign extension in library calls sign extension in switch (these all came from the FSF, were posted on the net, or were done here and sent back to the FSF, so should be in 2.7.0). The 1.42 version of gcc compiles much faster (about 3x; more if you are very short on physical memory, simply because the compiler is that much smaller) than version 2.6.3. The code is not quite as good, but many of the improvements in gcc 2.x are targeted towards RISC machines, and have less of an effect on the 386 architecture than on the SPARC, HP-PA, etc. We also disable by default some optimizations that work on machines with FPUs but not on machines without them. As others have mentioned before on this list, the reason is that our kernel does not emulate some complex instructions (such as FSIN and FCOS). The math library routines get the right answer, if a little slower. If you want to get the special instructions, use `-mfancy-math-387', and be sure you have a hardware FPU. -- In-Real-Life: Chris Torek, Berkeley Software Design Inc Berkeley, CA Domain: torek@bsdi.com +1 510 549 1145 -- With best regards -- Andrew Stesin. +380 (44) 2760188 +380 (44) 2713457 +380 (44) 2713560 "...Good marketing beats good technology, every time". Pity.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199510161429.QAA26398>