<GitHub58>
[artiq] whitequark pushed 1 new commit to new-py2llvm: http://git.io/vYJ9o
<GitHub58>
artiq/new-py2llvm dff4ce7 whitequark: Return LLVM IR module from LLVMIRGenerator.process.
ylamarre has joined #m-labs
<cr1901_modern>
Those running mibuild programmers on Linux: Do you have to become root in order to use, say xc3sprog, OpenOCD, etc?
chiggs has joined #m-labs
<rjo>
cr1901_modern: no. you usually muck wit udev to give yourself permissions on the device.
<cr1901_modern>
that's what I was afraid of. No udev equivalent on BSD
<cr1901_modern>
(AFAIK)
<rjo>
but you do have devices, iirc
<rjo>
and permissions
<rjo>
this is posix after all
<cr1901_modern>
Yes, I do have those. I guess I may have to change device permissions for /dev/ugen*
<cr1901_modern>
Use case is to create a C programmer for Mercury platform, since the ppl who made them thought C# was a good idea
<rjo>
.... TAG+="uaccess" should see much more widespread use.
<cr1901_modern>
libftdi calls libusb which on NetBSD opens /dev/ugen*
<cr1901_modern>
I think I'm just going to have to concede that using FPGA's on BSD is not straightforward unless you're familiar with *nix.
<rjo>
yes. set permissions. same thing on *nix. that can not be an uncommon occurence under *BSD.
<rjo>
don't you have good old devfs?
<cr1901_modern>
Don't think so. I know where my device lives anyway, just reading over some man pages
<cr1901_modern>
1. Need Linux emulation for ISE. 2. Need to recompile the kernel so that FTDI drivers attach to /dev/ugen* instead of /dev/ttyU* (Who the hell thought that was a good default?!). 3. Set /dev/ugen* permissions to 666
<rjo>
can't you symlink the device.
<rjo>
also 660 and root:plugdev or alike seems to be the consensus.
ohama has joined #m-labs
<cr1901_modern>
No, because the kernel will attach a different driver (uftdi), which is accessed through /dev/ttyU*
<rjo>
or create the ttyU* as well.
<rjo>
linux can dynamically detach drivers.
* cr1901_modern
misses the days where hardware wasn't plug and pray, tbh
<cr1901_modern>
My hesitation for setting permissions on /dev/ugen* is twofold: 1. ugen* is the Generic USB device. Every single endpoint is given a separate device file. That's quite a few device files I need to switch permissions manually :P
<cr1901_modern>
2. I have no idea if I attach my dev board in the future whether the same /dev/ugen[0-9].* will be used
<cr1901_modern>
rjo: In any case, you're right that the proper way on BSD is to just set the permissions manually. I just want to make sure that I do it right
<cr1901_modern>
rjo 660 and root "colon p" lugdev (the "colon p" became an emoji for me)?
sb0 has joined #m-labs
<GitHub14>
[artiq] sbourdeauducq pushed 6 new commits to master: http://git.io/vYUL8
<GitHub14>
artiq/master 179ca36 Sebastien Bourdeauducq: gui: basic scan support
<GitHub67>
[misoc] enjoy-digital pushed 1 new commit to master: http://git.io/vYtpN
<GitHub67>
misoc/master dfc207a Florent Kermarrec: litepcie: use data instead of dat in dma_layout (allow use of migen.actorlib.packet modules on dma dataflow)