lynx-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: lynx-dev Latest May lynx can cause AMD CPU to segfault when compi


From: vtailor
Subject: Re: lynx-dev Latest May lynx can cause AMD CPU to segfault when compi
Date: Mon, 03 Jun 2002 04:05:14 -0400

>     * From: address@hidden
>     * Date: Fri, 31 May 2002 14:13:38 -0400
>>     * Date: Tue, 28 May 2002 21:22:38 -0400
>>If you are using the MingW32 compile of the recent May lynx source, you
>>may start getting segfaults in crtdll.dll.  A quick fix that stops the
>>segfaults is to change all the "-O3 -mpentium" compiler options in
>>makelynx.bat to "-O".
>>
>>You may also want to try out the define that gives you an estimated time
>>of arrival "ETA" for downloaded files, and check out the LYUtils.c file
>>for a missing #ifdef that causes the MingW32 compile to misfire.
>
>P. S.   Other optimizations that don't cause crtdll.dll to crash are
>"-O5 -mpentium", "-O -mpentium", "-O5", etc.  Using "O5 -mpentium"
>causes a noticeable speed increase in lynx.exe operation.

The mystery deepens somewhat:  I downloades the last January lynx source,
and it does _not_ cause the compiler to crash crtdll.dll.  The conclusion
is that something changed in the source code since January that causes
MingW32 gcc to crash for -O3 -mpentium, and this change appears in all
the relevant .c files.  I have one candidate for the culprit: Extraneous,
but non-visible control characters that creep in with the use of certain
text editors.
>
>I should mention that this is a "Morgan" (disfigured Athlon XP) AMD
>processor at 1000 Mhz., but I think the problem should also occur with
>the previous "spitfire" series.  People using Linux on an AMD system
>might want to change the optimization settings on their Makefile files
>to determine whether the problem is repeatable under Linux.
>
>Also, this motherboard is an "all in one" motherboard, with the PCI
>video card interrupt tested at "on" and "off" (leave it "off" for best
>onbard video performance).  And, the L2 cache is disabled on this
>motherboard, because leaving it enabled doesn't improve benchmark
>performance or gcc compile speed, and disabling L2 lowers the operating
>CPU temperature by several degrees fahrenheit.

; To UNSUBSCRIBE: Send "unsubscribe lynx-dev" to address@hidden

reply via email to

[Prev in Thread] Current Thread [Next in Thread]