IRC channel logs

2020-07-28.log

back to list of logs

***Server sets mode: +nt
***nckx is now known as ^
***^ is now known as nckx
<damo22>youpi: it doesnt seem to like MACH_PORT_NULL for dotdot
<damo22>(ipc/send) invalid destination port.
<damo22> err = fsys_getroot (tc, MACH_PORT_NULL, MACH_MSG_TYPE_COPY_SEND,
<damo22> idlist, 3, idlist, 3, 0,
<damo22> &retry, retry_name, &root);
<damo22>does tc need to be the send right for machdev?
<damo22>or the control port handed from fsys_init
<damo22>does dotdot need to be something like a port for a dummy node?
<damo22>to hold the root fs
<youpi>See:
<youpi>utils/settrans.c: err = fsys_getroot (active_control,
<youpi>utils/settrans.c- MACH_PORT_NULL, MACH_MSG_TYPE_COPY_SEND,
<youpi>so a null port should be fine
<youpi>the dest of the rpc has to be the control port
<youpi>as provided by fsys_startup
<youpi>there is no "underlying" node inside rumpdisk for ext2fs to get mounted on, so dotdot really needs to be null
<damo22> part:2:device:/dev/wd0 / ext2fs writable,store-type=typed 0 0
<damo22> part:2:device:/dev/wd0 /dev/rump ext2fs writable,store-type=typed 0 0
<damo22> https://paste.debian.net/plain/1158050/
<gnu_srs>Seen this many times in gdb: A bug? warning: Can't wait for pid 1013: No child processes
<damo22>startup: notifying ext2fs part:2:device:/dev/wd0 of reboot... its getting stuck in diskfs_S_startup_dosync()
<damo22># ./testlock
<damo22>HEHE!! Locked
<damo22>IO ports are locked by rumpdisk
<damo22>hmm
<damo22>no wonder my networking is down
<damo22>wb youpi, i think ive almost got it working, it just hangs in diskfs_S_startup_dosync
<damo22>but after that times out, rumpdisk shuts down cleanly
<damo22> https://paste.debian.net/plain/1158050/ rumpdisk shutdown notification works
<youpi>yay :)
<youpi>just need to see why it would fail :)
<youpi>(I mean, time out)
<damo22>i think its the mutex locks
<damo22>in diskfs_S_startup_dosync
<damo22>something is causing it to not sync or something
<damo22>because it never exits that function
<youpi>that's surprising: it usually just works
<youpi>donc a sync call at the shell work without hanging?
<youpi>diskfs_S_startup_dosync is not doing much more than that
<damo22>i think that hangs too from memory
<youpi>that doesn't usually happen
<damo22>could it be io ports locked?
<youpi>I don't see why, the sync is not much more than the usual periodical sync
<damo22>i think my ahci controller uses the same ports as part of PCI CFG2 range in qemu iirc
<youpi>anyway, better put mach_prints to make sure where exactly it hangs
<youpi>not a problem, ext2fs doesn't do any hardware I/O
<damo22>hm ok
<damo22>i dont know why my rumpdisk install_as_translator seems like a no-op maybe im using the wrong port, anyway i have sent in some more patches, no hurry, am going to bed now