Re: gnumach - Page fault error
On Thu, Nov 09, 2017 at 10:34:33AM +0100, Samuel Thibault wrote: > Jens Arm, on jeu. 09 nov. 2017 08:56:21 +0100, wrote: > > Sorry for the late answer. > > This fix fixes the crashes. > > Cool :) Nice catch :). -- Richard Braun
Re: gnumach - Page fault error
Jens Arm, on jeu. 09 nov. 2017 08:56:21 +0100, wrote: > Sorry for the late answer. > This fix fixes the crashes. Cool :) Samuel
Re: gnumach - Page fault error
Hi Samuel Sorry for the late answer. This fix fixes the crashes. I can install hurd without crash and can install later many packages. Jens On Wed, 1 Nov 2017 02:07:52 +0100 Samuel Thibault wrote: > Jens Arm, on mer. 04 oct. 2017 10:39:42 +0200, wrote: > > Sorry this change fixes nothing, I have retried it and I got the same trace > > as you can see in gnumachError.jpg attachment. > > Ok. I have built a new image with the directmap fix, could you try it? > > http://people.debian.org/~sthibault/tmp/mini.iso > > Samuel >
Re: gnumach - Page fault error
Jens Arm, on mer. 04 oct. 2017 10:39:42 +0200, wrote: > Sorry this change fixes nothing, I have retried it and I got the same trace > as you can see in gnumachError.jpg attachment. Ok. I have built a new image with the directmap fix, could you try it? http://people.debian.org/~sthibault/tmp/mini.iso Samuel
Re: gnumach - Page fault error
Hello, Richard Braun, on mar. 03 oct. 2017 10:54:12 +0200, wrote: > On Tue, Oct 03, 2017 at 12:14:39AM +0200, Samuel Thibault wrote: > > Jens Arm, on lun. 02 oct. 2017 19:20:57 +0200, wrote: > > > Here is the trace to the error in the attached PNG. > > > Is is reproduced with the mini.iso from Samuel. > > > The trace shown after I type "trace" is the same as shown automatically. > > > > Ok, it looks like traces I have already seen in our buildd VMs. It > > means the IDE driver is trying to push data to disk some memory which > > doesn't exist any more. I was wondering whether perhaps it's just that > > the IDE layer misses putting some reference to make sure that the memory > > doesn't get swapped out while being written to disk. > > > > Richard, does that ring a bell? > > It's possible. I didn't work much on the IDE driver because it already > did what was required for highmem support, but I may have been wrong. > The code in linux/dev/glue/block.c:device_write uses vm_map_enter with > a NULL object, which normally prevents paging, That said, in block.c's rdwr_full, one can read bh->b_data = (char *) phystokv(pmap_extract (vm_map_pmap (device_io_map), (((vm_offset_t) *buf) + (nb << bshift; AIUI that can't work if the pages are in highmem, thus the issue (actually unrelated to eviction). I'll have a look at using a bounce buffer there. Samuel
Re: gnumach - Page fault error
Hello, Richard Braun, on mar. 03 oct. 2017 10:54:12 +0200, wrote: > I see some scary code in vm_map_enter that may merge the new mapping > with another with a valid object. Mmm. I have disabled it and rebuilt an image. Jens, could you try this? http://people.debian.org/~sthibault/tmp/mini.iso Samuel
Re: gnumach - Page fault error
On Tue, Oct 03, 2017 at 12:14:39AM +0200, Samuel Thibault wrote: > Jens Arm, on lun. 02 oct. 2017 19:20:57 +0200, wrote: > > Here is the trace to the error in the attached PNG. > > Is is reproduced with the mini.iso from Samuel. > > The trace shown after I type "trace" is the same as shown automatically. > > Ok, it looks like traces I have already seen in our buildd VMs. It > means the IDE driver is trying to push data to disk some memory which > doesn't exist any more. I was wondering whether perhaps it's just that > the IDE layer misses putting some reference to make sure that the memory > doesn't get swapped out while being written to disk. > > Richard, does that ring a bell? It's possible. I didn't work much on the IDE driver because it already did what was required for highmem support, but I may have been wrong. The code in linux/dev/glue/block.c:device_write uses vm_map_enter with a NULL object, which normally prevents paging, but I see some scary code in vm_map_enter that may merge the new mapping with another with a valid object. I wonder if map entry merging should just be disabled completely, as it wouldn't be the first time it caused problems, and isn't that useful considering how unoptimized the VM system is in general, and that scalable data structures are now used. I don't have time to look at the details, but I could see a mistake right at the beginning of linux/dev/glue/block.c:device_write, where the local vm_map_copy object may be discarded while still uninitialized. Probably unrelated but still noteworthy. -- Richard Braun
Re: gnumach - Page fault error
Hello, Jens Arm, on lun. 02 oct. 2017 19:20:57 +0200, wrote: > Here is the trace to the error in the attached PNG. > Is is reproduced with the mini.iso from Samuel. > The trace shown after I type "trace" is the same as shown automatically. Ok, it looks like traces I have already seen in our buildd VMs. It means the IDE driver is trying to push data to disk some memory which doesn't exist any more. I was wondering whether perhaps it's just that the IDE layer misses putting some reference to make sure that the memory doesn't get swapped out while being written to disk. Richard, does that ring a bell? Samuel
Re: gnumach - Page fault error
Jens Arm, on dim. 01 oct. 2017 17:30:04 +0200, wrote: >Not Found >The requested URL /~sthibault/hurd-i386/tmp/mini.iso was not found on this > server. >Apache Server at people.debian.org Port 443 Ergl, sorry, that was meant to be https://people.debian.org/~sthibault/tmp/mini.iso Samuel
Re: gnumach - Page fault error
Hi Samuel OK, I will try to reproduce the error with your iso. But I can't download it: Not Found The requested URL /~sthibault/hurd-i386/tmp/mini.iso was not found on this server. Apache Server at people.debian.org Port 443 Jens On Sun, 1 Oct 2017 16:57:58 +0200 Samuel Thibault wrote: > Re, > > Jens Arm, on jeu. 28 sept. 2017 09:12:58 +0200, wrote: > > The error is shown only a few seconds then it reboots automatically, so I > > had to be very quick taking this photo. > > Perhaps this photo helps to fix this. > > Probably not, unfortunately, because they don't provide source lines. I > have however uploaded a mini.iso with debugging symbols enabled (ftr, > gnumach version 2:1.8+git20170911-1) on > > https://people.debian.org/~sthibault/hurd-i386/tmp/mini.iso > > When it crashes, you should be able to type > > trace > > to get a complete backtrace which would be useful. > > Samuel -- Jens Arm
Re: gnumach - Page fault error
Re, Jens Arm, on jeu. 28 sept. 2017 09:12:58 +0200, wrote: > The error is shown only a few seconds then it reboots automatically, so I had > to be very quick taking this photo. > Perhaps this photo helps to fix this. Probably not, unfortunately, because they don't provide source lines. I have however uploaded a mini.iso with debugging symbols enabled (ftr, gnumach version 2:1.8+git20170911-1) on https://people.debian.org/~sthibault/hurd-i386/tmp/mini.iso When it crashes, you should be able to type trace to get a complete backtrace which would be useful. Samuel
Re: gnumach - Page fault error
Hello, Jens Arm, on jeu. 28 sept. 2017 09:12:58 +0200, wrote: > I have tried to eMail Richard and Justus because they have worked in the past > on the VM, > but I can't reach them. The eMail returning to me. Please always use mailing lists, such as bug-hurd@gnu.org . Mailing only some random people will always only get random answers. Mailing the whole list will get the minimal response time. > I wanted to give DebianHurd a try on my old Acer NetBook with IntelAtom and > 1G Ram. > This DebianHurd-ISO was used for the test: > > https://d-i.debian.org/daily-images/hurd-i386/20170927-00:42/netboot/mini.iso Perhaps just for getting more information you could try other images, such as from http://ftp.ports.debian.org/debian-ports-cd/hurd-i386/current/ https://people.debian.org/~sthibault/hurd-i386/installer/cdimage/daily/ > The attached PNG shows a photo of a reproducible error showing up on > different locations while installing. > Sometimes while installing the base system and sometimes when installing the > additional packages after tasksel. > The error is shown only a few seconds then it reboots automatically, so I had > to be very quick taking this photo. > Perhaps this photo helps to fix this. Samuel