Tuesday, 2022-06-14

*** tpb <[email protected]> has joined #openrisc00:00
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)00:57
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc01:08
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)02:03
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc02:10
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)02:42
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc02:44
*** littlebo1eep <littlebo1eep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc02:57
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)03:01
*** littlebo1eep <littlebo1eep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Remote host closed the connection)04:01
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc04:06
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)04:40
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc05:04
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)06:29
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc06:46
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Remote host closed the connection)08:30
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc08:35
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)09:25
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc09:25
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)09:35
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc09:35
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)10:27
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc10:30
*** littlebo1eep <littlebo1eep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc13:47
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)13:47
*** littlebo1eep <littlebo1eep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)14:12
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc14:18
*** littlebo1eep <littlebo1eep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc15:01
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)15:02
*** tpb_ <[email protected]> has joined #openrisc15:18
*** tpb <[email protected]> has quit IRC (Killed (NickServ (GHOST command used by tpb_)))15:18
*** tpb_ is now known as tpb15:18
*** littlebo1eep <littlebo1eep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)15:29
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc15:35
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)15:39
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc15:49
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)16:20
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc16:21
*** littlebo1eep <littlebo1eep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc16:26
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)16:28
*** littlebo1eep <littlebo1eep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)17:54
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc17:54
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)18:02
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc18:06
*** shorne <[email protected]> has quit IRC (*.net *.split)18:18
*** shorne <[email protected]> has joined #openrisc18:21
*** shorne <[email protected]> has quit IRC (Ping timeout: 246 seconds)18:23
*** shorne <[email protected]> has joined #openrisc18:25
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)19:06
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc19:15
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Remote host closed the connection)20:15
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc20:27
shornezx2c4: alright, after sorting out some interrupt and timer issues the mttcg patches are looking stable21:28
shornebut.. running accel tcg,thread=single is not working well when I set to 4 cores21:29
shorneit's maybe something you could explain21:30
shorneall cores are waiting on the same spinlock21:30
shorne 0xc045c578 in _get_random_bytes (buf=0xc10918f4, len=4) at drivers/char/random.c:36421:30
shornethen..21:31
shorne0xc045c318 in crng_make_state (chacha_state=0xc1091850, random_data=0xc10918f4 "", random_data_len=4) at include/linux/spinlock.h:32421:31
shornesorry, maybe the whole trace is better21:31
shornedeetail21:38
shornehttps://gist.github.com/stffrdhrn/b7d726c3bc74cd7732d70465fefb161821:38
zx2c4shorne: not sure i understand what youre reporting... Single threaded multicore tcg always has horrible performance because it just round robins cpus, right? So a spin lock requires a full round robin in the worst case22:18
zx2c4shorne: OH!! NO22:18
zx2c4What youre hitting is22:19
zx2c4https://lore.kernel.org/all/[email protected]/22:19
tpbTitle: [PATCH] locking/lockdep: Use sched_clock() for random numbers. - Sebastian Andrzej Siewior (at lore.kernel.org)22:19
zx2c4shorne: apply that patch22:19
zx2c4https://git.kernel.org/tip/4051a81774d6d8e28192742c26999d6f29bc0e6822:22
tpbTitle: kernel/git/tip/tip.git - Unnamed repository; edit this file 'description' to name the repository. (at git.kernel.org)22:22
shornezx2c4: yup boots with that one22:22
shornethanks, I had lockdep and all debug stuff on for debugging22:23
zx2c4Whew22:23
zx2c4It's in a tip urgent branch so hopefully it'll hit rc322:23
shornecool, I thought you might know it22:24
shornein that case my patches might be ready to go22:25
shorneIll do a bit more testing, i still see some instability in places22:25
shornealso, I think the 200 second clock jumps might be fixed22:25
zx2c4Im headed home from playing a show now. When im back ill adjust my CI patch to use the virt machine and try building your github branch and sew22:25
zx2c4See22:25
zx2c4Oh cool  what was the 200 sec root cause?22:25
shornewell, I am not sure, so I maybe didn't fix it, but during boot or1k tries to synchronize cpu timers between cpus, when it does that if there is a big skew during the update it will case a jump22:28
shorneI changed how it is initialized in qemu to try to avoid big skews22:28
shornebut it might not be the only issue22:28
shornethe 200sec constant is not explained by that22:28
shornezx2c4: what do you play?22:29
zx2c4Guitar (jazz)22:29
zx2c4I thought i saw the jump with -smp 1 though22:29
shornecool, I play guitar, wish I knew my theory better to play jazz22:30
shorneyeah, thats right it does happen with smp 1, but the initialization is wrong even with smp 122:30
shornei.e. the last_clk (which is system time) was always initialized to zero, so on the first clock update the time jumps from 0 to current time nanos, and then it uses that calculation to set the tick delta22:31
shorneI updated it so last_clk initializes to the system start time nanos22:32
zx2c4Ahhh makes sense22:32
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has quit IRC (Ping timeout: 240 seconds)23:12
zx2c4shorne: which branch do i want to play with?23:16
zx2c4or1k-virt-2?23:16
zx2c4hm i dont see the mttcg patches here23:18
zx2c4nor the rng-seed thing23:18
*** littlebobeep <littlebobeep!~alMalsamo@gateway/tor-sasl/almalsamo> has joined #openrisc23:19
zx2c4shorne: looks like reboot isnt working with your or1k-virt-2 branch23:28
zx2c4[    2.532000] reboot: Restarting system23:28
zx2c4[63620.380000] Reboot failed -- System halted23:28
shornezx2c4: sorry I dont' have the latest patches submitted yet23:31
shorneplease give a day or so, I am at work and need to commit everything23:31
shorneI removed semihosting, so we need to be sure to have the syscon driver enabled on the linux side23:32
shorneenabling syscon should be added to the virt_defconfig, I may not have committed that yet too23:33
zx2c4ahh23:35
shornezx2c4: ok pushed the qemu changes here: https://github.com/stffrdhrn/qemu/commits/or1k-virt-223:47
shornewill work on kernel patches, I found one issue with the or1k interrupt controller driver23:47

Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!