<sb0>
cr1901_modern, ok but the problem is on the openocd side, right?
<sb0>
there was a similar issue with ISE
<cr1901_modern>
let me check something first
<sb0>
is that TCL? you can use {} to disable escaping
<cr1901_modern>
You are correct, it's OpenOCD's fault lol
<cr1901_modern>
i.e. the script as printed by Python3 preserves the backslashes (which makes sense)
<cr1901_modern>
Adding braces to disable escaping doesn't work either. So whoops :P
<cr1901_modern>
mithro: Will do.
<sb0>
really?
<cr1901_modern>
Well based on my testing, it still strips out all the escape chars
<sb0>
$ tclsh8.5
<sb0>
% puts {a\b}
<sb0>
a\b
<cr1901_modern>
Hmmm... interesting
balrog has quit [Ping timeout: 260 seconds]
<sb0>
are you escaping { and } correctly for python's "".format()?
<cr1901_modern>
No, I wasn't. Fixed.
<cr1901_modern>
And now it works
balrog has joined #m-labs
Mon_1 has quit [Quit: This computer has gone to sleep]
ylamarre has joined #m-labs
<sb0>
cr1901_modern, can you send the patch?
<cr1901_modern>
Sure, in a few minutes. Also found a new issue while testing OpenOCD flash; fpga_program is no longer a valid command
<cr1901_modern>
far as I can tell, the name changed from something generic to FPGA-family specific between the time that rjo submitted a patch to add jtagspi support and the time his changes were actually accepted
Guest44455 has quit [Quit: This computer has gone to sleep]
Mon_1 has joined #m-labs
Mon_1 has quit [Quit: This computer has gone to sleep]
Mon_1 has joined #m-labs
<cr1901_modern>
sb0: Patch sent. However, the problem with fpga_program remains. I'll need to wait until I can talk to rjo.
balrog has quit [Ping timeout: 260 seconds]
balrog has joined #m-labs
<sb0>
cr1901_modern, thanks
<GitHub16>
[migen] sbourdeauducq pushed 1 new commit to master: http://git.io/vRCm0
<GitHub16>
migen/master daa53ef William D. Jones: Prevent backslashes in (Windows) paths from being escaped by OpenOCD's TCL implementation.
<sb0>
cr1901_modern, what happens with fpga_program? I couldn't get openocd to work either ...
<cr1901_modern>
fpga_program isn't a valid command; in between the time that rjo submitted the patch that introduced the command, and when it was placed into the main repository, it seems that the name changed
<cr1901_modern>
And now it's device-specific xc6s_program :/
Mon_1 has quit [Quit: This computer has gone to sleep]
<rjo>
yeah. i made it device specific because it is.
<rjo>
but i see that it makes generic scripts a bit harder to write.
<sb0>
cr1901_modern, ah, it fails before that here. i can't even open the jtag cable.
<rjo>
the good thing is that in principle it allows having multiple devices on the chain.
<cr1901_modern>
sb0: You have a minispartan6+?
<rjo>
maybe one should submit a patch that sill exposes a default fpga_program (and maybe the name is not perfect) for the first device on the chain.
<sb0>
cr1901_modern, no, tried (summarily) with other boards...
<cr1901_modern>
mithro: Are you able to confirm that flashing the minispartan6+ works? I can get a verification that the flash is programmed, but it won't load the bitstream when power is cycled... :?
<sb0>
rjo, do you have any experiments to test the new compiler with btw?
<cr1901_modern>
rjo: Neither of those keep their configuration after power cycle for me
<sb0>
I'm trying out RStudio right now which seems to have a docking system much better than pyqtgraph's....
<sb0>
I notice that they package the full Qt libraries with it, which I totally understand now considering how many bugs the damn thing has
balrog has quit [Ping timeout: 245 seconds]
balrog has joined #m-labs
sb0 has quit [Read error: Connection reset by peer]
sb0 has joined #m-labs
<sb0>
hmm, in fact, not really
<sb0>
rjo, so what is the decision regarding CSRConstant deriving from fhdl.structure.Constant vs. having the Constant as attribute?
<rjo>
sb0: i would have to write an experiment. but that is basically still the same request that we had a year ago, right? ;)
<rjo>
cr1901_modern: they are not supposed to. this is for CI testing.
<rjo>
sb0: i think my last patch set made the Constant a child of CSRConstant. even if it gets ugly then (CSRConstant.value.value etc).
ylamarre has quit [Quit: ylamarre]
<sb0>
ah yes, it's in your repository...
<rjo>
do you want me to send them?
<rjo>
sb0: i have the artiq patch for that as well, just need to fix a few conflicts it.
<rjo>
s/it././
<sb0>
rjo, do you use bits_sign?
<rjo>
where?
<rjo>
artiq or misoc/migen
<sb0>
CSRConstant.bits_sign in misoc
<sb0>
doesn't it make sense to have CSRConstant derive from _CSRBase and merge CSRConstant.bits_sign with _CSRBase.size?
<rjo>
if Constant has it, so should CSRConstant.
<rjo>
byr _CSRBase.size has no sign afaik
<rjo>
but
<sb0>
indeed, but if we support signed CSRConstants we might as well support signed other-CSRs as well
<rjo>
i want signed constants for redpid.
<sb0>
on the gateware side it's easy, just create the signals with bits_sign afaict
<sb0>
the C routines will need some trickery for sign extension
<rjo>
the header templating is not the most beautiful thing in misoc.
<sb0>
yeah... what would you propose instead?
<rjo>
for the signedness thing: i would bundle "merging CSRConstant.bits_sign with _CSRBase.size" with "making _CSRBase.size signed"
<rjo>
for templating: maybe real templating with one of the python templating engines is in order.
<sb0>
making _CSRBase.size signed?
<sb0>
rjo, ah yes, the artiq patch doesn't apply with the new runtime/compiler...
<rjo>
"supporting signed CSRs in general"
<sb0>
well, yes
<sb0>
anyway, small detail. the misoc patches are ok
<sb0>
rjo, can you send a refreshed artiq patch and i will merge this?
<sb0>
rjo, I assume you have tested it?
<rjo>
if test-compiled the artiq stuff. not tested on hardware
<rjo>
i will have some time next week for testing and artiq in general.
Mon_ has joined #m-labs
Mon_ is now known as Guest22692
Guest22692 has quit [Client Quit]
Mon_1 has joined #m-labs
Mon_1 has quit [Quit: This computer has gone to sleep]
Mon_1 has joined #m-labs
Mon_1 has quit [Quit: This computer has gone to sleep]
stekern has quit [Ping timeout: 250 seconds]
<GitHub146>
[misoc] sbourdeauducq pushed 7 new commits to master: http://git.io/vRWlz
<GitHub146>
misoc/master beb25ab Robert Jordens: csr: add CSRConstant and related support
<GitHub146>
misoc/master 44c1fcf Robert Jordens: add_constant -> Config...
<GitHub146>
misoc/master b75ef9c Robert Jordens: csr/Config: remove add_constant; Config is no Module
<GitHub64>
[artiq] sbourdeauducq pushed 2 new commits to master: http://git.io/vRWlg
<GitHub64>
artiq/master 5db1f97 Robert Jordens: top.add_constant() -> top.config[] (CSRConstant)...
<GitHub64>
artiq/master 7886827 Robert Jordens: CSRConstant: also port DDS constants
stekern has joined #m-labs
sb0 has quit [Read error: Connection reset by peer]
sb0 has joined #m-labs
wbx has quit [Ping timeout: 260 seconds]
wbx has joined #m-labs
stekern has quit [Ping timeout: 260 seconds]
stekern has joined #m-labs
FabM has quit [Quit: ChatZilla 0.9.92 [Firefox 42.0/20151029151421]]
wbx has quit [Ping timeout: 260 seconds]
wbx has joined #m-labs
stekern has quit [Ping timeout: 245 seconds]
stekern has joined #m-labs
sb0 has quit [Read error: Connection reset by peer]
Mon_1 has joined #m-labs
mindrunner has quit [Quit: quit]
mindrunner has joined #m-labs
Mon_1 has quit [Quit: This computer has gone to sleep]
Mon_ has joined #m-labs
Mon_ is now known as Guest93542
Guest93542 has quit [Client Quit]
<cr1901_modern>
sb0: What's the current stance on external dependencies for MiSoC? You may like pyexpander for creating header templates
<cr1901_modern>
It's like m4, but you have access to "whatever's currently in scope" in your Python environment as well.
<cr1901_modern>
(m4 would work too, but I'm certainly not going to assume that Windows users have it. I kinda wish there wasn't a gmake dependency too.)
Mon_ has joined #m-labs
Mon_ is now known as Guest11006
Guest11006 has quit [Quit: This computer has gone to sleep]
Mon_ has joined #m-labs
Mon_ is now known as Guest90444
Gurty has joined #m-labs
Gurty has joined #m-labs
Guest90444 has quit [Quit: This computer has gone to sleep]