On Thu, Aug 15, 2002 at 05:27:53PM +0200, Juliusz Chroboczek wrote: > Yep. Try adding ``Option "PciRetry" "true"'' to your Device section. > Upstream is trying to be benchmark- rather than user-friendly. > > (To expand on that: the XFree86 4 client scheduler specially optimises > the case of a single client. That only happens when benchmarking.) Hmph. Well, as a package maintainer for a distribution, I care a lot more about users having machines that don't lock up than I do about benchmarks. Do you happen to know where I can reverse this default in the X server? Hmm, actually, looking at the sources, I don't see anyplace where this option defaults to "on" on a server-wide basis. Only specific drivers appear to be cognizant of this option and, typically, they don't even consistently call it by the same name internally: drivers/siliconmotion/smi_driver.c: pSmi->NoPCIRetry = FALSE; drivers/trident/trident_driver.c: pTrident->UsePCIRetry = FALSE; /* Not Supported */ So which upstream is the problem? The Trident driver author? -- G. Branden Robinson | Psychology is really biology. Debian GNU/Linux | Biology is really chemistry. email@example.com | Chemistry is really physics. http://people.debian.org/~branden/ | Physics is really math.
Description: PGP signature