IRC channel logs

2023-01-27.log

back to list of logs

<damo22> EFL=00000046 [---Z-P-] HLT=1
<damo22>sometimes the cpu gets into a halted state with interrupts disabled
<damo22>it will never wake up right?
<damo22>both cpus :
<damo22>EIP=c1028361 EFL=00000246 [---Z-P-] CPL=0 II=0 A20=1 SMM=0 HLT=1
<damo22>i think the first task always dies
<damo22>and it just sits in idle_thread
<damo22>feat-smp2-hangs
<damo22>we're getting close
<damo22>TLB(1>0)
<damo22>start acpi:
<damo22>db{0}> show all tasks
<damo22> ID TASK NAME [THREADS]
<damo22> 0 f5906ea0 gnumach [10]
<damo22> 1 f5906d00 pci-arbiter [1]
<damo22> 2 f5906c30 rumpdisk [1]
<damo22> 3 f5906b60 ext2fs [1]
<damo22> 4 f5906a90 exec [1]
<damo22>acpi is missing
<damo22>LAPIC timer configured on cpu5
<damo22>BSP: Completed SMP init
<damo22>it boots intermittently with -smp 2
<damo22>some race condition
<Pellescours>on my side It start to boot but rumpdisk loose a lot of interruptions with apic, so I can’t boot until the end
<damo22>i can push serialised smp init
<damo22>it inits with -smp 6 but doesnt quite boot
<damo22>pushed
<damo22>Pellescours: in ioapic.c there is a critical timer problem, set calibrate_ticks to hz as the multiplier
<damo22>not 16 or 10
<damo22>it booted with -smp 4 but using 400% load
<damo22>top - 22:34:06 up 5 min, 1 user, load average: 6.82, 6.68, 6.91
<damo22>but nothing is using the cpus much
<damo22>i just pushed my latest commit one liner
<damo22>it boots/hangs every few boots
<damo22>Pellescours: it should work with gnumach disk driver
<damo22>+ apic
<Pellescours>niice
<damo22>something is wrong because all cores are very busy
<damo22>but top says nothing much is running except load is 6
<damo22>also console is stuck
<damo22>but i can ssh to it
<Pellescours>and if you try to use the cpu a lot do you see perfs?
<damo22>Thread 1 received signal SIGQUIT, Quit.
<damo22>0xc100054c in smp_pmap_update (apic_id=4129009312) at ../i386/i386/smp.c:56
<damo22> https://paste.debian.net/plain/1268631
<damo22>check this out, the state of the processors when it boots with -smp 4
<damo22>Pellescours: i dont think i can use the APs because they are processing nothing
<damo22>they must be looping or something
<damo22>i can barely log in
<damo22>cpus are going nuts and my fan is going beserk
<damo22>Pellescours: how is yours looking? i am kind of stuck
<Pellescours>I did tested only with rump and because this irq lost for piixide I did not tested much more
<Pellescours>Oh I kept my linux image, switching to it and testing
<Pellescours>damo22: I cloned your repo, autoreconf -fi, configure --enable-acpi --enable-ncpus=4, make gnumach.gz and it says db_on is in undef-bad
<damo22>--enable-apic
<Pellescours>it’s enable-apic I put
<damo22>git checkout feat-smp2-hangs
<Pellescours>on it
<Pellescours>Ah I know, I didn’t enabled kdb
<Pellescours>But this means that the build is "broken" if kernel debugger is not enabled. I don’t know if it’s your branch only or in master too
<Pellescours>damo22: I have "hd0: irq timeout: status=0.50 { DriveReady SeekComplete }
<damo22>ok
<damo22>i dont know sorry, its late im going to sleep
<Pellescours>no problem. So the db_on problem is only on your branch (not on master). And it seems that the apic is not ready yet. So it’s blocking smp