Re: X problems & 5.0... -RELEASE?
* De: Marc Recht <[EMAIL PROTECTED]> [ Data: 2002-10-19 ] [ Subjecte: Re: X problems & 5.0... -RELEASE? ] > > Note that during these, mp3s keep playing, gtk-gnutella keeps downloading > > things, etc., it seems X just isn't updating the display... Even a resize > But, it's not always so. For me most of the time everything stops. > Including playing mp3s/oggs, resizing windows and so on. And after some > time just restarts/continues.. If you're using an X-driven mouse (i.e. not sysmouse) and a GUI mp3 player (I am using madplay), I imagine this would be so... I'm talking about things either queueing up, or continuing fine, while X is retarded. > > of a sizable gtk frame I did while it was frozen took affect when it > > unfroze... Have you tried loggign into a system during such a freeze to > > see if it's sleeping or such? I would, except ENOSPAREBOX and if I try > > to switch to a console the screen getsd this black-and-green wooshy colour and > > then it sticks there. > Switching to the console also doesn't work for me in these situations. > But I can't login via SSH and reboot the box. > > Marc > > -- > "Premature optimization is the root of all evil." -- Donald E. Knuth -- Juli Mallett <[EMAIL PROTECTED]> | FreeBSD: The Power To Serve Will break world for fulltime employment. | finger [EMAIL PROTECTED] http://people.FreeBSD.org/~jmallett/ | Support my FreeBSD hacking! To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
> Note that during these, mp3s keep playing, gtk-gnutella keeps downloading > things, etc., it seems X just isn't updating the display... Even a resize But, it's not always so. For me most of the time everything stops. Including playing mp3s/oggs, resizing windows and so on. And after some time just restarts/continues.. > of a sizable gtk frame I did while it was frozen took affect when it > unfroze... Have you tried loggign into a system during such a freeze to > see if it's sleeping or such? I would, except ENOSPAREBOX and if I try > to switch to a console the screen getsd this black-and-green wooshy colour and > then it sticks there. Switching to the console also doesn't work for me in these situations. But I can't login via SSH and reboot the box. Marc -- "Premature optimization is the root of all evil." -- Donald E. Knuth signature.asc Description: This is a digitally signed message part
Re: X problems & 5.0... -RELEASE?
On Sat, 19 Oct 2002, Juli Mallett wrote: > Note that during these, mp3s keep playing, gtk-gnutella keeps downloading > things, etc., it seems X just isn't updating the display... Even a resize > of a sizable gtk frame I did while it was frozen took affect when it > unfroze... Have you tried loggign into a system during such a freeze to > see if it's sleeping or such? I would, except ENOSPAREBOX and if I try > to switch to a console the screen getsd this black-and-green wooshy colour and > then it sticks there. I have... The XFree86 process was runing wildly. Was able to kill it and regain the system. Did not think to gdb anything.. Truss coredumps trying to attach to it. -- Hi! I'm a .signature virus! Copy me into your ~/.signature to help me spread! To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
* De: Eric Anholt <[EMAIL PROTECTED]> [ Data: 2002-10-19 ] [ Subjecte: Re: X problems & 5.0... -RELEASE? ] > >From the reports: > > Both stable and current users get the "self-healing" hang, where the X > server responds to nothing but the mouse moves, and at some minutes > later time it continues and responds to those actions. One person said > they'd had this since at least current in July. Note that during these, mp3s keep playing, gtk-gnutella keeps downloading things, etc., it seems X just isn't updating the display... Even a resize of a sizable gtk frame I did while it was frozen took affect when it unfroze... Have you tried loggign into a system during such a freeze to see if it's sleeping or such? I would, except ENOSPAREBOX and if I try to switch to a console the screen getsd this black-and-green wooshy colour and then it sticks there. juli. -- Juli Mallett <[EMAIL PROTECTED]> | FreeBSD: The Power To Serve Will break world for fulltime employment. | finger [EMAIL PROTECTED] http://people.FreeBSD.org/~jmallett/ | Support my FreeBSD hacking! To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
On Sun, 2002-10-13 at 23:00, Eric Anholt wrote: > On Sun, 2002-10-13 at 21:14, Kris Kennaway wrote: > > On Sun, Oct 13, 2002 at 11:28:51PM -0400, Wesley Morgan wrote: > > > > > I know there is some work being done on the recent signal changes to fix > > > some things, but are we sure this is the problem? I would hate to see > > > release schedules pushed back because these problems are lost in the > > > noise, and I can't see a release being made that has a known unstable X. > > > > I thought this was believed to be a bug in X that was exposed by > > kernel changes. > > > > Kris > Could anyone who is having stability issues with X please email me > privately if they are using either -current before September or > -stable? If not, without some sort of hints of where an issue really > is, I'm going to chalk this up to kernel bugs. Just to let people know what's going on with this on my end: I've got my laptop up to a fresh kernel, world, and X as of 10/17 or so. I've got a reproducible X server crash with XFree86 + glxgears alone (DRI disabled). I'm working on getting backtraces to see if anything useful can be produced. However, gdb521 is crashing if I start XFree86 from it (gdbing that gdb produced only silliness -- gbs exiting semicleanly or senseless backtraces). gdb521 can attach to a running XFree86 fine apparently, but then it doesn't get the module info. On my -stable box, gdb521 appears to start XFree86 fine, but on stable (and current iirc) ^Cing in gdb results in nothing happening and needing to kill the gdb or the XFree86 because they go unresponsive. If I can get gdb52 to be useful, I'll add a patch to XFree86-4-Server (and dri-devel maybe?) to compile debuggable X Server/modules and install them properly. >From the reports: Both stable and current users get the "self-healing" hang, where the X server responds to nothing but the mouse moves, and at some minutes later time it continues and responds to those actions. One person said they'd had this since at least current in July. Folks with kernels later than a couple weeks ago get X crashes all the time. Updated world wasn't necessary to get it (in my case), updated world+kernel didn't help (others), and updated world+kernel+X didn't help (my case, too). Note that just about any X crash will result in a signal six reported by the kernel, because one X crash causes another while it tries to recover (reset to the console, etc), and on the second crash that gets caught it aborts. To see what started the mess, look at the console output from startx if you used startx. You're looking for the first "Fatal error" -- later stuff is trying to recover from that crash that got caught. If you use xdm, it's in your /var/log/xdm-errors iirc. If you blame this on type1/bezier, make sure you actually have an error message about bezier or something else in your log before the abort. All of the type1 module's aborts have a reason printed before the abort. -- Eric Anholt <[EMAIL PROTECTED]> http://people.freebsd.org/~anholt/dri/ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
Tim Robbins writes: > > It's worth noting that 4.7-RELEASE (w/ gcc 2.95.4) fails *more* test cases > than -CURRENT when -O is used to compile paranoia. Does 2.95.4 fail the same tests or different ones? Also, this program doesn't deal with signals, which is where I suspect the problem may be (ie, not the compiler but the kernel). If you understand this code, can you write and install signal handler for, say, sigusr1 and have another programe fire sigusr1's at the math test? Drew To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
On Mon, 14 Oct 2002, Steve Kargl wrote: > On Tue, Oct 15, 2002 at 12:39:26PM +1000, Tim Robbins wrote: > > On Mon, Oct 14, 2002 at 01:00:46PM -0700, Kris Kennaway wrote: > > > > [...] > > > Did anyone test -current with the various FP test suites people posted > > > about last week? I run some of them occasionally (last ran ucbtest on June 18), but they are perfectly irrelevant to this problem since they don't use signals in any significant way. > > Yes. I ran paranoia from http://cm.bell-labs.com/netlib/paranoia/ and > > found that FP arithmethic is satisfactory when -O is not used, and no > > -march or -mcpu options are used. > > > > However, compiling with -O causes a lot of failures. > > Are you using an intel cpu? You need to add -ffloat-store > to get the correct results due to the 80 bit FPU registers. > Otherwise, intermediate results are stored with higher > precision. This is a well-known bug in gcc. Intermediate results may be computed with higher precision. paranioa knows this and uses lots of assignments to clip the results to double (or single) precision, aa is required to work by C standards (very fuzzily in C90 and very clearly in C99). This doesn't actually work in the i386 gcc. The -ffloat-store hack works around this and gives much the same pessimizations as would a standards conforming compiler. Bruce To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
On Tue, Oct 15, 2002 at 12:39:26PM +1000, Tim Robbins wrote: > On Mon, Oct 14, 2002 at 01:00:46PM -0700, Kris Kennaway wrote: > > [...] > > Did anyone test -current with the various FP test suites people posted > > about last week? > > Yes. I ran paranoia from http://cm.bell-labs.com/netlib/paranoia/ and > found that FP arithmethic is satisfactory when -O is not used, and no > -march or -mcpu options are used. > > However, compiling with -O causes a lot of failures. Are you using an intel cpu? You need to add -ffloat-store to get the correct results due to the 80 bit FPU registers. Otherwise, intermediate results are stored with higher precision. Both of the following commands: f77 -o a -ffloat-store dpara.f f77 -o a -ffloat-store -O dpara.f yield No failures, defects nor flaws have been discovered. Rounding appears to conform to the proposed IEEE standard P754 except possibly for Double Rounding during Gradual Underflow. The arithmetic diagnosed appears to be Excellent! End of Test. Otherwise, optimization yields your results. > > Here are the messages: > > Seeking Underflow thresholds UfThold and E0. > DEFECT: Difference underflows at a higher threshold than products. > ... > Can `Z = -Y' overflow? > Trying it on Y = -inf . > finds a FLAW: -(-Y) differs from Y. > ... > FAILURE: Comparisons involving +--inf, +-inf > and +-4.94066e-324 are confused by Overflow. > ... > DEFECT: Badly unbalanced range; UfThold * V = -inf > is too far from 1. > > SERIOUS DEFECT:X / X differs from 1 when X = -inf > instead, X / X - 1/2 - 1/2 = nan . > > > The summary message: > > The number of FAILUREs encountered = 1. > The number of SERIOUS DEFECTs discovered = 1. > The number of DEFECTs discovered = 2. > The number of FLAWs discovered = 1. > > The arithmetic diagnosed has unacceptable Serious Defects. > Potentially fatal FAILURE may have spoiled this program's subsequent diagnoses. -- Steve To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
On Mon, Oct 14, 2002 at 01:00:46PM -0700, Kris Kennaway wrote: [...] > Did anyone test -current with the various FP test suites people posted > about last week? Yes. I ran paranoia from http://cm.bell-labs.com/netlib/paranoia/ and found that FP arithmethic is satisfactory when -O is not used, and no -march or -mcpu options are used. However, compiling with -O causes a lot of failures. Here are the messages: Seeking Underflow thresholds UfThold and E0. DEFECT: Difference underflows at a higher threshold than products. ... Can `Z = -Y' overflow? Trying it on Y = -inf . finds a FLAW: -(-Y) differs from Y. ... FAILURE: Comparisons involving +--inf, +-inf and +-4.94066e-324 are confused by Overflow. ... DEFECT: Badly unbalanced range; UfThold * V = -inf is too far from 1. SERIOUS DEFECT:X / X differs from 1 when X = -inf instead, X / X - 1/2 - 1/2 = nan . The summary message: The number of FAILUREs encountered = 1. The number of SERIOUS DEFECTs discovered = 1. The number of DEFECTs discovered = 2. The number of FLAWs discovered = 1. The arithmetic diagnosed has unacceptable Serious Defects. Potentially fatal FAILURE may have spoiled this program's subsequent diagnoses. It's worth noting that 4.7-RELEASE (w/ gcc 2.95.4) fails *more* test cases than -CURRENT when -O is used to compile paranoia. f77 -O seems to also generate bad code for dpara.f, the FORTRAN version of paranoia. Tim To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
Kris Kennaway writes: > > I think we're all waiting for Julian and Jonathan Mini to fix this.. > > It would probably help if they had some sort of a test program that > > could duplicate the bug in a controlled setting without a lot of > > confusing application software running.. > > Did anyone test -current with the various FP test suites people posted > about last week? I couldn't make any sense of them. Perhaps the gcc regression tests might be better. Where does one download them? Drew To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
On Mon, Oct 14, 2002 at 01:58:26PM -0400, Andrew Gallatin wrote: > > Nate Lawson writes: > > On Mon, 14 Oct 2002, Andrew Gallatin wrote: > > > Given that it (apparently) happens on X servers compiled months ago, > > > my gut feeling is that it is a bug in the kernel floating point > > > context save/restore in the presence of signals. > > > > > > I still can't run X without crashing the *kernel* because of the > > > conglomeration of hacks that was added to i386/machdep.c to paper-over > > > floatingpoing problems after the latest KSE brea^W import. The > > > last machdep.c that works for me is 1.539. This has been dragging > > > on for nearly 2 weeks. > > > > > > Drew > > > > I don't know how to say this strongly enough but can someone PLEASE PLEASE > > fix this properly? It is preventing real work from getting done. > > I think we're all waiting for Julian and Jonathan Mini to fix this.. > It would probably help if they had some sort of a test program that > could duplicate the bug in a controlled setting without a lot of > confusing application software running.. Did anyone test -current with the various FP test suites people posted about last week? Kris msg44658/pgp0.pgp Description: PGP signature
Re: X problems & 5.0... -RELEASE?
Nate Lawson writes: > On Mon, 14 Oct 2002, Andrew Gallatin wrote: > > Given that it (apparently) happens on X servers compiled months ago, > > my gut feeling is that it is a bug in the kernel floating point > > context save/restore in the presence of signals. > > > > I still can't run X without crashing the *kernel* because of the > > conglomeration of hacks that was added to i386/machdep.c to paper-over > > floatingpoing problems after the latest KSE brea^W import. The > > last machdep.c that works for me is 1.539. This has been dragging > > on for nearly 2 weeks. > > > > Drew > > I don't know how to say this strongly enough but can someone PLEASE PLEASE > fix this properly? It is preventing real work from getting done. I think we're all waiting for Julian and Jonathan Mini to fix this.. It would probably help if they had some sort of a test program that could duplicate the bug in a controlled setting without a lot of confusing application software running.. Drew To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
On Mon, 14 Oct 2002, Andrew Gallatin wrote: > Given that it (apparently) happens on X servers compiled months ago, > my gut feeling is that it is a bug in the kernel floating point > context save/restore in the presence of signals. > > I still can't run X without crashing the *kernel* because of the > conglomeration of hacks that was added to i386/machdep.c to paper-over > floatingpoing problems after the latest KSE brea^W import. The > last machdep.c that works for me is 1.539. This has been dragging > on for nearly 2 weeks. > > Drew I don't know how to say this strongly enough but can someone PLEASE PLEASE fix this properly? It is preventing real work from getting done. Thanks, Nate To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
On 13 Oct 2002 23:00:08 -0700 Eric Anholt <[EMAIL PROTECTED]> wrote: > Could anyone who is having stability issues with X please email me > privately if they are using either -current before September or > -stable? If not, without some sort of hints of where an issue really > is, I'm going to chalk this up to kernel bugs. Running -current as of Oct 8, X4 as of May 15. I hadn't the problem with a kernel from one or two months ago. Maxim was trying to find the date where it breaks, don't know how far he is. I only see signal 6, regardless of the loading of the type1 module for X. It only aborts if I have my MUA running (see headers). No problem with e.g. Galeon. I also see those hangs, but wasn't able to find something specific to trigger them. But at every temporary hang the mouse pointer doesn't freeze, and the mouse clicks get played back at the correct positions after X unfreezes. I'm also able to freeze the system hard just by killing mldonkey (CVS version), but I don't know if this is related, as the mouse pointer isn't movable, I don't think it is related, but who knows. Bye, Alexander. -- To boldly go where I surely don't belong. http://www.Leidinger.net Alexander @ Leidinger.net GPG fingerprint = C518 BC70 E67F 143F BE91 3365 79E2 9C60 B006 3FE7 To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
Kris Kennaway writes: > On Sun, Oct 13, 2002 at 11:28:51PM -0400, Wesley Morgan wrote: > > > I know there is some work being done on the recent signal changes to fix > > some things, but are we sure this is the problem? I would hate to see > > release schedules pushed back because these problems are lost in the > > noise, and I can't see a release being made that has a known unstable X. > > I thought this was believed to be a bug in X that was exposed by > kernel changes. Given that it (apparently) happens on X servers compiled months ago, my gut feeling is that it is a bug in the kernel floating point context save/restore in the presence of signals. I still can't run X without crashing the *kernel* because of the conglomeration of hacks that was added to i386/machdep.c to paper-over floatingpoing problems after the latest KSE brea^W import. The last machdep.c that works for me is 1.539. This has been dragging on for nearly 2 weeks. Drew To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
On Sun, 2002-10-13 at 21:14, Kris Kennaway wrote: > On Sun, Oct 13, 2002 at 11:28:51PM -0400, Wesley Morgan wrote: > > > I know there is some work being done on the recent signal changes to fix > > some things, but are we sure this is the problem? I would hate to see > > release schedules pushed back because these problems are lost in the > > noise, and I can't see a release being made that has a known unstable X. > > I thought this was believed to be a bug in X that was exposed by > kernel changes. > > Kris I've heard some people saying that it's the bezier bug. When I had moved to late September/early October kernels, I saw sig11s and the hangs (some temporary, some resulting in reset switch before they could be temporary), but never that message people have mentioned about the Beziers. I couldn't see *any* pattern to my crashes. It often happened while I was reading email, but then I spend a decent amount of time reading mail. The [EMAIL PROTECTED] archived message mentioned previously about Type 1 issues listed two bugs. One was an abort on an error, which we aren't experiencing as far as I've heard. The other was not failing requests for very large fonts, which shouldn't be happening too often and shouldn't have anything to do with the kernel version. Could anyone who is having stability issues with X please email me privately if they are using either -current before September or -stable? If not, without some sort of hints of where an issue really is, I'm going to chalk this up to kernel bugs. -- Eric Anholt <[EMAIL PROTECTED]> http://people.freebsd.org/~anholt/dri/ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
Kris Kennaway wrote: > On Sun, Oct 13, 2002 at 11:28:51PM -0400, Wesley Morgan wrote: > > I know there is some work being done on the recent signal changes to fix > > some things, but are we sure this is the problem? I would hate to see > > release schedules pushed back because these problems are lost in the > > noise, and I can't see a release being made that has a known unstable X. > > I thought this was believed to be a bug in X that was exposed by > kernel changes. That's an incredibly funny thought... There are a lot of things you could characterise as "bugs exposed by kernel changes" that involve replacing Linux with FreeBSD. Or vice versa. 8-). -- Terry To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
On Sun, 13 Oct 2002, Kris Kennaway wrote: > On Sun, Oct 13, 2002 at 11:28:51PM -0400, Wesley Morgan wrote: > > > I know there is some work being done on the recent signal changes to fix > > some things, but are we sure this is the problem? I would hate to see > > release schedules pushed back because these problems are lost in the > > noise, and I can't see a release being made that has a known unstable X. > > I thought this was believed to be a bug in X that was exposed by > kernel changes. The Type1/bezier problems supposedly are, but as for the rest... I don't know. That's what my concern is -- if we uncovered a bug, even though its "an X problem", the OS will still be blamed. Could the X server be doing something so absolutely completely braindamaged that these new-fangled signal things cause it to simply quit working? Surely an X bug of this magnitude would not be so localized and would have turned up on other platforms and even 4.x. I have a lot of faith in the RE team, and faith in 5.0 being a great new branch... It has some features that are a MUST for desktops and laptops -- firewire, acpi, cardbus, to name a few -- but a stable X is also a must. We don't want to become like Apache 2 ;) -- Hi! I'm a .signature virus! Copy me into your ~/.signature to help me spread! To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: X problems & 5.0... -RELEASE?
On Sun, Oct 13, 2002 at 11:28:51PM -0400, Wesley Morgan wrote: > I know there is some work being done on the recent signal changes to fix > some things, but are we sure this is the problem? I would hate to see > release schedules pushed back because these problems are lost in the > noise, and I can't see a release being made that has a known unstable X. I thought this was believed to be a bug in X that was exposed by kernel changes. Kris msg44602/pgp0.pgp Description: PGP signature
X problems & 5.0... -RELEASE?
The last tentative date for 5.0-RELEASE I have seen is "late November, early December"... This does seem odd since I haven't seen a DP2, but... Quite a few people seem to be having serious problems with XFree86 for 3-4 weeks, everything from sig 6's, the "bezier" crashes, to strange freezes that sometimes "correct" themselves. The bezier bug is somehow related to a problem with the Type1 module, and disabling it helps a little. I have rebuilt my kernel, world, QT, KDE, and X system with-mpentiumpro instead of pentium3. I have not yet seen any SIGABRT's, but I am getting unrecoverable crashes (console stuck) and the "freezes" that will pass after 5-10 minutes of nail-biting wait. During those freezes, I have logged in via the network and attempted to use ktrace and truss on the wildly out-of-control XFree86 process. truss dumps core (probably a thread issue?) and ktrace generates no output. I'm going to try to rebuild with debugging symbols to attach with GDB, which was successful but produced no meaningful output. The "temporary freeze" seems to always occurr in the Konqueror location bar when it attempts to complete a URL as I key it in, but the runaway process is XFree86. I know there is some work being done on the recent signal changes to fix some things, but are we sure this is the problem? I would hate to see release schedules pushed back because these problems are lost in the noise, and I can't see a release being made that has a known unstable X. Thanks WNM -- Hi! I'm a .signature virus! Copy me into your ~/.signature to help me spread! To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
