adding a few more people from peng/kernel team to comment

Could this also be causing https://buganizer.corp.google.com/issues/34394866?

It could be. When the network didn't come up, the system seemed to give up after a while and go back to normal ChromeOS. I could then restart the app manually, and it would run fine. I don't have to do that once I made sure the firmware blob was available.

One thing that seems different, though, is that I pretty clearly saw mention of the firmware blob it wanted in the logs as the system came up, and I didn't see mention of that anywhere in the other bug. The network did eventually come up, but there was a delay of 45 seconds to a minute.

https://chromium-review.googlesource.com/#/c/430272/
https://chromium-review.googlesource.com/#/c/430231/

Moch or YH - would one of you be able to drive the landing of this FW?

Gabe, I see both CLs in c#6 abandoned. Are new/modified CLs in the works?

No new CLs are in the works.

Any reason the previous CLs were abandoned? Can we just reuse those CLs - we can add the necessary LICENSE file.

I found out that I won't actually be able to join the team, and so stopped the work I was doing. Feel free to reuse the CLs.

Since I'll be moonlighting with the team for a little while, I'll work on getting this checked in. We do still need to figure out how far spread it should be, ie panther vs. guado vs. guado-cfm.

I wonder whether the machine I have is actually a panther, or if it's a guado who's firmware just claims that it's a panther. If it actually is a guado, it could be that things work without this firmware blob (off hand, I think it might), and this is actually just a firmware bug where the firmware reports the wrong type of device.

HWID?

hardware_id: PANTHER G5F-34A-A6T
fwid: Google_Panther.4920.24.26

The guado image I'm using for other purposes seems to have a similar message, except that it wants rtl_nic/rtl8168g-2.fw (-2 instead of -1). It doesn't seem to get stuck like the panther image, but this is probably still worth fixing there too.

Definitely looks like a Panther device.

Gabe, reassigning to you. lmk if I can help in anyway.

Well, we do still need to decide conclusively which machines we want to install that firmware on. Could you please take a look at CLs listed above and weigh in? I think at this point I'm fairly confident that that change is ok and that we want that blob on all panthers, but confirmation would be nice and I do still need a +2.

I'll put together some equivalent CLs for guado and add you as a reviewer.

The blob should be included on all panthers (there was only a single realtek ethernet controller used across all SKUs)

would be nice to understand how this was working otherwise, but if we don't think including/using the firmware is a problem, maybe it's not worth figuring out considering the age of the device

when did this issue start showing up ?  in  issue 682578 , we've made some changes related to the udev upgrade that caused large delays on devices like lumpy where firmware was unavailable.  could you try the latest canary image and see if that works for panther too ?

It is believed that these "stuck for 1 minute" reports are all related to the udev change, and mka@ is on top of it.

Separately:

> if we don't think including/using the firmware is a problem, maybe it's not worth figuring out considering the age of the device

We have a question pending with Realtek about this.  Some zako devices are having intermittent, irreproducible wired connectivity issues (b/34394866).  We're wondering if there are any known errata that would be fixed by bundling the firmware blob.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/overlays/board-overlays/+/eb3bbd7667a0b3feaa11a35708636e8810c12af4

commit eb3bbd7667a0b3feaa11a35708636e8810c12af4
Author: Gabe Black <gabeblack@chromium.org>
Date: Wed Mar 01 09:26:47 2017

panther: Turn on the rtl_nic/rtl8168g-1.fw firmware blob.

Signed-off-by: gabeblack@google.com
BUG= chromium:682445 
TEST=Used cros deploy to update the linux-firmware package on panther.
Rebooted it and saw errors related to the firmware blob were gone, and the
network came up without a significant delay.

Change-Id: I2510f272e61097ec6aecc497bbdd7ffb8f2a759a
Reviewed-on: https://chromium-review.googlesource.com/430272
Commit-Ready: Gabe Black <gabeblack@chromium.org>
Tested-by: Gabe Black <gabeblack@chromium.org>
Reviewed-by: Abhishek Bhardwaj <abhishekbh@google.com>
Reviewed-by: Mohammed Habibulla <moch@google.com>
Reviewed-by: Mike Frysinger <vapier@chromium.org>

[modify] https://crrev.com/eb3bbd7667a0b3feaa11a35708636e8810c12af4/overlay-panther/make.conf

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/fe66eab2744f0a074f190f5f0089a2169494cda6

commit fe66eab2744f0a074f190f5f0089a2169494cda6
Author: Gabe Black <gabeblack@chromium.org>
Date: Wed Mar 01 13:04:05 2017

kernel: Enable installing the rtl_nic/rtl8168g-1.fw firmware blob.

This blob is installed if the linux_firmware_rtl8168g-1 USE flag is set.

BUG= chromium:682445 
TEST=Used the linux_firmware_rtl8168g-1 USE flag to enable the firmware blob
on panther, and verified that it was installed successfully and the kernel on
that device could load it.
Signed-off-by: gabeblack@google.com

Change-Id: I672268c200209d29c3cfc33cb0bca3dc59681a1a
Reviewed-on: https://chromium-review.googlesource.com/430231
Commit-Ready: Gabe Black <gabeblack@chromium.org>
Tested-by: Gabe Black <gabeblack@chromium.org>
Reviewed-by: Mike Frysinger <vapier@chromium.org>

[modify] https://crrev.com/fe66eab2744f0a074f190f5f0089a2169494cda6/sys-kernel/linux-firmware/linux-firmware-9999.ebuild
[add] https://crrev.com/fe66eab2744f0a074f190f5f0089a2169494cda6/licenses/LICENCE.rtl_nic

The panther part of this should be resolved now. Other bugs can be filed for other systems, specifically guado and zako.