02:38 <sb0> sturmflut2, android kernel still crashes here

06:15 <sb0> ysionneau, did you get minicon to work with ddr? it might be a good idea to use it with ddr3 on the kc705 as well ...

07:23 <ysionneau> sb0: not yet

08:39 <ysionneau> sb0: what was the reason to add "set -e" to artiq_flash.sh script to make it fail at the first encountered error?

08:39 <sb0> yes

08:39 <ysionneau> I mean, why make it fail like that?

08:40 <sb0> unfortunately, it doesn't work very well because xc3sprog doesn't set a different status code when it fails

08:40 <ysionneau> yep :/

08:40 <sb0> to help prevent the script print "done" when something has gone wrong

08:40 <ysionneau> hum ok

08:41 <ysionneau> for instance when I forget to put -t pipistrello (and therefore it choses the cable for kc705) then the first xc3sprog fails and the script just returns without printing anything

08:41 <ysionneau> maybe I can put the set -e later, right before the first flashing

08:41 <sb0> yes, that's a xc3sprog problem, it should return a different status code

08:42 <sb0> set -e will catch other errors, unfortunately not the xc3sprog ones (which are also the most common...)

08:44 <ysionneau> what I don't get is why the "trap ERR" was not enough and set -e was needed

08:44 <ysionneau> since in the trap I do exit 1

08:45 <ysionneau> trap is supposed to run a handler if some error that might have caused a set -e script to abort happens

08:46 <sb0> er, doesn't trap deal with signals, no status code?

08:47 <sb0> "trap is a shell builtin which executes the command when the shell receives signal"

08:47 <ysionneau> trap <function_name> ERR will call function_name in case of a command failure

08:47 <ysionneau> A SIGNAL_SPEC of ERR means to execute ARG each time a command's failure would cause the shell to exit when the -e option is enabled.

08:48 <sb0> ah: "The KornShell uses an ERR trap that is triggered whenever set −e would cause an exit. This is allowable as an extension, but was not mandated, as other shells have not used it."

08:49 <sb0> looks like a pretty obscure feature. i'd stick with "set -e" only.

08:49 <ysionneau> bash has this feature at least

08:49 <sb0> also, the "set -e" should be the first thing in the script

08:50 <ysionneau> but not sh

08:50 <ysionneau> set -e being the first thing in the script makes some of the error checking in the script useless

08:50 <ysionneau> for instance the message that warns that maybe there is issue to access usb device

08:50 <ysionneau> and tells to use the udev script

08:51 <ysionneau> I would put the set -e at least right after this

08:51 <ysionneau> (in place of the trap ERR)

08:52 <ysionneau> this way you don't get the "Done." but you can still get some useful warning

08:53 <ysionneau> (I mean you don't get the "Done." when there is a flashing error)

08:53 <sb0> without set -e at the beginning, ARTIQ_PREFIX can be silently set wrong (or not set) if there is a problem with the artiq or python install, for example

08:54 <ysionneau> ah, right

08:55 <ysionneau> ok let's keep it this way then :(

08:55 <ysionneau> too bad trap ERR is not standard

08:56 <ysionneau> I'll clean it up and remove the trap then

08:56 <sb0> maybe you can do set +e/set -e around manual error checks for status codes

08:57 <ysionneau> ah yes, indeed

08:57 <sb0> but well. shell is a lousy programming language.

08:58 <ysionneau> it's horrible

08:58 <ysionneau> writing a really *portable* shell script is a nightmare

09:01 <ysionneau> sb0: since with mkfs we can generate a flashable file with the FS, I can add a -f FILE option to artiq_flash.sh to flash the fs. In fact I already did it in my local tree for testing purposes I can just push it if it's OK

09:04 <sb0> artiq_mkfs is a stopgap solution because your fs_write did not work. eventually, the flash filesystem should be accessible remotely, without jtag

09:04 <ysionneau> yep I'm doing the implementation of the remote access right now

09:05 <ysionneau> ok so no need for the flash file solution then

09:12 <GitHub194> [artiq] fallen pushed 1 new commit to master: http://git.io/vUyO5

09:12 <GitHub194> artiq/master e9b166b Yann Sionneau: artiq_flash.sh: some cleanup

09:54 <mindrunner> how can I write byte by byte with a DMAWriter? If I try to write single bytes, I always end up in having every 8th byte modified in my memory? Is that expected? I am using a slightly modified version of the memtest module

11:01 <travis-ci> fallen/artiq#161 (flash_storage_for_sb - 036b27b : Yann Sionneau): The build has errored.

11:01 <travis-ci> Build details : http://travis-ci.org/fallen/artiq/builds/62533556