IRC logs for #openrisc Sunday, 2017-05-28

--- Log opened Sun May 28 00:00:04 2017
shorneHello, its been quiet lately06:28
shorneI have been working on gdb06:28
shornea few things06:28
shorne 1. moved out the target decsription patch, as dicussed with yao qi on the gdb list (its more of a general gdb patch)06:29
shorne(todo on that is to build a testsuite for the feature to allow arbitrary groups in the xml target descriptor06:30
shorne2. cleaned up the sim testsuite indentation/fsf code style06:30
shorne3. right now I am working on fixing a lots of TODOs that have been left in the sim testsuite06:31
shorneI seems there is one issue with the cpu description06:31
shorneAnyone have and idea on this?06:31
shornedni (l-mul) "l.mul reg/reg/reg"06:32
shorne          ((MACH ORBIS-MACHS))06:32
shorne          ("l.mul $rD,$rA,$rB")06:32
shorne          (+ OPC_ALU rD rA rB (f-resv-10-7 #x30) OPC_ALU_REGREG_MUL)06:32
shorne          (sequence ()06:32
shorne                    (sequence ()06:32
shorne                              ; 2's complement overflow06:32
shorne                              (set BI sys-sr-ov (mul-o2flag WI rA rB))06:32
shorne                              ; 1's complement overflow06:32
shorne                              (set BI sys-sr-cy (mul-o1flag WI rA rB))06:32
shorne                              (set rD (mul WI rA rB))06:32
shorne                              )06:32
shorne                    (if (andif sys-sr-ov sys-sr-ove)06:32
shorne                        (raise-except...06:32
shorneBut... in the spec it says06:32
shornerD[31:0] ← rA[31:0] * rB[31:0]06:32
shorneSR[OV]   ← signed overflow06:32
shorneNothing about setting carry bit06:33
shornethis test is failing, anyone else agree the cpu description (lisp) needs fixing?06:34
shorneI suggest just remove (set BI sys-sr-cy (mul-o1flag WI rA rB))08:41
-!- ZipCPU_ is now known as ZipCPU08:48
shornehmm, similar with divide09:10
shorne          "l.div $rD,$rA,$rB"09:11
shorne          (+ OPC_ALU rD rA rB (f-resv-10-7 #x30) OPC_ALU_REGREG_DIV)09:11
shorne          (sequence ()09:11
shorne                    (if (ne rB 0)09:11
shorne                        (sequence ()09:11
shorne                                  (set BI sys-sr-cy 0)09:11
shorne                                  (set WI rD (div WI rA rB))09:11
shorne                                  )09:11
shorne                        (set BI sys-sr-cy 1)09:11
shorne                        )09:11
shornethe spec says09:11
shornerD[31:0] ← rA[31:0] / rB[31:0]09:11
shorne← rB[31:0] == 009:11
shorne(lisp) sets carry, but spec says to set overflow09:11
shorneI dont think many even care09:12
windelFor those interested in my bare metal example from yesterday, I put it here:
windelThe only question that I still have is whether the MMU must be configured before startup11:12
wbxrth: hi. are you going to push openrisc gcc support upstream?12:13
rthwbx: no.  as i understand it we don't have assignments from all the people that contributed.12:14
wbxrth: so it is based on the same "problematic" code? damn.12:14
rthwbx: there has been no concerted effort to either (1) rewrite from scratch or (2) verify that the problematic bits are gone.12:15
wbxrth: may I ask an unrelated alpha architecture question?12:23
wbxi have read an email about __NR_oldumount hanlding.
wbxrth: why the change was never made into upstream kernel?12:24
rthheh.  i don't remember.  maybe i thought michael was going to take care of it.12:25
rthhe has done more than i in the last decade...12:26
wbxrth: i playing a little with alpha - uClibc-ng. static system works. but with porting I got again into the issue with the special umount handling.12:26
wbxrth: on which hardware you run linux? i got recently a DEC multia box and trying to netboot via mopd. 4.9.30 kernel does not boot. what do you think is a safe kernel version to get something going?12:27
rthwbx: I have no hardware anymore.  I use qemu.12:27
rthi think i'm currently running 4.10, but i'd have to check.12:28
shornewindel: keeps logging off, not sure if they check the irc log online, but I dont think the mmu needs to be enabled.18:49
shorneThe code is compiked using "api.asm()" which is a python compiler18:50
shorneok, I see pure python compiler18:53
shorneIf you are having problems you can try running qemu with trace enabled TRACE="-D trace.txt -d in_asm,exec,int,cpu,op_opt"18:54
--- Log closed Mon May 29 00:00:05 2017

Generated by 2.15.2 by Marius Gedminas - find it at!