00:07ccr: santa checks his list via 'top'?
00:08alanc: he knows if you are sleeping, he knows when you're on the CPU
00:16dcbaker: but not the GPU apparently
00:16alyssa: all I want for Christmas is gputop for mali
01:38robclark: alyssa: btw drm/msm's $debugfs/perf is a fairly simple/crude way to measure how busy the gpu is.. perfcntrs are ofc better
18:04imirkin: i'm trying to do a dry-run release with the latest xorg-modular, and i'm seeing:
18:04imirkin: Error: the path "/srv/xorg.freedesktop.org/archive/individual/driver" on the web server does not exist.
18:04imirkin: Error: processing module "." failed.
18:05imirkin: any thoughts?
18:11imirkin: oh, looks like it's not detecting that it's nouveau properly
18:12imirkin: ah no, nevermind. need to set USER_NAME
18:29imirkin: how do people pgp-sign the announcement emails?
18:30imirkin: i'm seeing some detached some signatures, but it's unclear exactly which part of the email they're signing
18:46dcbaker: You should be pgp signing the body and putting it in a multipart
18:46dcbaker: Your client should be able to do that for you
18:49imirkin: dcbaker: how would i do it with the gpg tool?
18:49imirkin: i plan on sending the email with git send-email
18:51robclark: imirkin: this looks like the script that I used to use to do that:
18:51dcbaker: Oh man, no idea. You're also asking me to remember the details of how pgp in email works... Lol. I think it's a detached signature, but then you create a multipart and put the payload in one part and the signature in the other
18:53imirkin: robclark: thanks!
18:53dcbaker: Working on an email client made me realize how much duct tape and bailing write email is. Hint: all of it
18:53imirkin: dcbaker: oh, i know all about that :)
18:53imirkin: robclark: ok, so that makes sense. you sign it without headers, and then include that as the body of the email
18:53imirkin: er, include the clearsign thing
18:54robclark: yeah, sounds about right.. one of the benefits of moving libdrm_freedreno into mesa is I haven't had to make a release for a *long* time ;-)
18:55robclark: but luckily I still had the script around
18:55robclark: I think maybe too much of my long-term memory exists in random little scripts scattered all over the place
19:12imirkin: robclark: the key is remembering the scripts' existence
19:12robclark: heheh, yeah
19:21imirkin: robclark: thanks, looks like it worked
19:21imirkin: or at least didn't fail very obviously
19:22imirkin: i'm sure the one person who verifies pgp signatures will pipe up to complain.
21:37Venemo: cwabbott or jekstrand, it seems to me that nir_shader_lower_instructions doesn't work with intrinsics that don't have a destination SSA def (such as stores), do we have something that works for this case?
21:57anholt: there's an MR for that, but I use nir_shader_instructions_pass
21:58Venemo: nice, that's almost what I need
21:58Venemo: anholt: which MR is that?
21:58anholt: don't have it off hand.
21:59Venemo: maybe this one? https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8177
22:22Venemo: Well, I think it's a good change so I added an r-b
22:22Venemo: Hope this is okay