Oh, Merry Christmas, Happy New Year, Happy Solstice, Happy Yule, and I
ferget the rest 'cause my family's christian an I don't really hear much
'bout the other holidays this time of year... but Happy Holidays no
matter -what- you celebrate! :)
Now on this message:
Prithee, anyone know what this's about?
This -is- the right area to ask AFAIK. I've heard of a bugposting site
but haven't found it... (I just do lynx 'n sometimes it's hard to find a
Anyways, I've been playing with the X code. Downloaded a new CVS last
night, compiled and ran. Guess what? (no changes to the code whatsoever,
going from standard CVS)
pciReadWord: Alignment Error: Cannot read 16 bits at offset 28.
Traced it down to a 'pciReadWord(<bus details>, PCI_PCI_BRIDGE_IO_REG)'
(line 987, xc/programs/Xserver/hw/xfree86/common/xf86bus.c,
calling into line 303+, xc/programs/Xserver/hw/xfree86/os-support/bus/Pci.c)
PCI_PCI_BRIDGE_IO_REG is defined as 0x1c in
<.../xfree86/common/xf86Pci.h>, line 323.
Can anyone explain this? This will -always- error out by my guessing.
Unless it's a problem in Pci.c. I don't get it.
Anyways, I disabled the error-checking and this happened. Logs are
included. I'm on a 14.4bps modem connection to the net and am running
Mandrake 6.0. I have no choice over this, the nearest linux retailer is
about 1000km away. Or maybe more, not sure if there's one in Edmonton but
probably. XFree86 was originally 3.3.1, I installed the originally
released rpm's of the binaries and they worked beautifully. Barring a few
bugs of course (hopefully since fixed). I need to see if latest CVS fixes
things but this doesn't help any *Sigh*. Incidentally, compiling (pick a
download of the DRI/X source) has always given me these problems. I've
got time to mess with it now though.... :)
Thanks and don't mind my rambling on but I hope it figures out in telling
me how to find the problem.
Thanks, and G'day, eh? :)
On Sun, Dec 26, 1999 at 09:18:26PM -0700, teunis wrote:
> This -is- the right area to ask AFAIK. I've heard of a bugposting site
> but haven't found it... (I just do lynx 'n sometimes it's hard to find a
Sure, you should ask questions here. We want to encourage people to post
bugs in the bug database so that they get tracked. Navigating with lynx
can be tough. Here's the URL:
> Anyways, I've been playing with the X code. Downloaded a new CVS last
> night, compiled and ran. Guess what? (no changes to the code whatsoever,
> going from standard CVS)
> Xalloc error: Tail corrupt in Xfree() for small block (adr=0x852a9c8, val=0x3b)
> Xalloc error: Tail corrupt in Xfree() for small block (adr=0x852a970, val=0x8537504)
> Xalloc warning: Xalloc(<0) ignored..
> Fatal server error:
> Caught signal 11. Server aborting
It looks like what you're getting here is actually a memory corruption
problem. It just happens that the bus management code is the unlucky
victim where the memory manager notices the problem. I haven't seen it,
so it is a bit track down, but I would take a look at the tdfx driver to
see if you can spot any malloc/free/overrun type errors. Otherwise,
3.9.17 is due out shortly and we'll be upgrading to that. A LOT of core
X code has been updated between 3.9.16 (which we're using in
dri.sourceforge.net) and 3.9.17, so if the problem was elsewhere it may
have been fixed.