<larsc> hmpf... something broke during rebase :/
<kristianpaul> i'll aply all my patches to my gps-sdr branch, as i get  lots of conflits when merging for last milkymist revision..
<kristianpaul> again.
<kristianpaul> oh,minimac changed lot stuff, now it have some ram and also conbus changed..
<kristianpaul> lekernel: so you go back to shared bus with minimac2, no more xbar..
<kristianpaul> okay i need to update some diagrams..
<kristianpaul> but why shared bus again? is it related that minimac have some cache now?
<kristianpaul> oh, minimac dont have dma anymore? lets confirm
<larsc> mwalle: btw what means "qemu: fatal: fetching nop sequence", can we cause it to trigger an exception on the guest instead of crashing qemu?
<kristianpaul> yes, minimac2 is slave now
<kristianpaul> good, at least now is simpler than before
<kristianpaul> .win 14
<CIA-48> milkymist: Sebastien Bourdeauducq master * r28248ed / (3 files): Updated SoC diagram (Cristian Paul Peñaranda Rojas) - http://bit.ly/hpeuFp
<CIA-48> flickernoise: Sebastien Bourdeauducq desktop * r80a7009 / (11 files): Desktop icons (slow) - http://bit.ly/ePojMv
<kristianpaul> oh no!, more libs ;-)
<kristianpaul> ah, keyboard binding sget lost after reboot..
<kristianpaul> cool feature :D
<kristianpaul> lekernel: by default all patches are influenced by both mic and line in, isnt?
<kristianpaul> s
<kristianpaul> wtf i did, now i'm botting in rescue mode..
<kristianpaul> now all is normal
<kristianpaul> after a powercycle and i just did that..
<kristianpaul> hmm is the automatic build tested...
<kristianpaul> 6837420 bytes sent in 78.24 secs (85.3 kB/s)
<kristianpaul> :D
<kristianpaul> 6837420 bytes received in 11.56 secs (577.5 kB/s)
<kristianpaul> :O
<kristianpaul> awesome, i even can upload/download stuff when milkdrop running
<kristianpaul> wtf page 37637 in gc has no object: 519 920 51
<kristianpaul> i was uploading a 12Mb file by ftp
<kristianpaul> i also got
<kristianpaul> netout: Broken pipe
<kristianpaul> 452 Error writing file.
<kristianpaul> he, nice nmap dint break rtems
<kristianpaul> and hping just drop some crc and package too short messages
<mwalle> larsc: there is some logic that halts the vm instead of cpu_abort, maybe i should put that there too, otoh i was asked to remove that nop detection
<mwalle> qemu guesses if the memory is not initialized, that is there are 5 consecutive zero opcodes (00 00 00 00, something with srui) and aborts
<larsc> mwalle: hm, i can see what goes wrong now. but i  wonder why it hasn't occurred before
<larsc> at the end of an syscall the stackpointer is always restored to the userspace stackframe
<larsc> this is a problem if we did an syscall from kernelspace, which obviously has no userspace stackframe