Bug 18830

Summary:	Missing iwlwifi microcode 7260-17
Product:	Mageia	Reporter:	Jérôme Hénin <heninj>
Component:	RPM Packages	Assignee:	Thomas Backlund <tmb>
Status:	RESOLVED FIXED	QA Contact:
Severity:	normal
Priority:	Normal	CC:	geiger.david68210
Version:	5
Target Milestone:	---
Hardware:	All
OS:	Linux
Whiteboard:
Source RPM:		CVE:
Status comment:

Description Jérôme Hénin 2016-07-01 11:16:37 CEST

On my machine the wifi driver prints the following error:
Direct firmware load for iwlwifi-7260-17.ucode failed with error -2

Then it defaults to version 16, but only after a minute, which is enough for the automatic connection to a known network to fail.

The kernel driver requests firmware version 17, but package iwlwifi-agn-ucode-20160516 only provides version 16.

This can be fixed by adding iwlwifi-7260-17.ucode, as committed there:

https://git.kernel.org/cgit/linux/kernel/git/iwlwifi/linux-firmware.git/commit/?id=66bac644550d9f68f6c2dd6228d79258962f0d67

David GEIGER 2016-07-01 11:19:15 CEST

CC: (none) => geiger.david68210
Assignee: bugsquad => tmb

Comment 1 Thomas Backlund 2016-07-01 11:33:48 CEST

Yes I know about the iwlwifi firmware repo (I tend to sync from it from time to time,) but its considered a "testing repo" and when Intel guys are happy with it, they send a pull request so it ends up in:

https://git.kernel.org/cgit/linux/kernel/git/firmware/linux-firmware.git/

Have you tested that the iwlwifi-7260-17.ucode actually works for you ?

Comment 2 Jérôme Hénin 2016-07-01 12:42:20 CEST

Yes, version 17 works well for me.

I am sure it's fine not to distribute version 17, but then the driver would need to be fixed so that it loads version 16 properly at boot.

Comment 3 Thomas Backlund 2016-07-01 13:03:43 CEST

ok, great.

as for the driver behaviour, that is by design...

most intel wireless drivers supports 3-4 driver apis, and by default they try to get the newest api first, and then trying older apis one at a time until it finds a matching firmware and starts that one...

But I have planned a new firmare update on mga5 for some other issues, so I'll pull in theese api 17 then too..

Comment 4 Jérôme Hénin 2016-07-01 13:09:58 CEST

I see. When another API comes out, this problem might occur again in the future.

The root cause is that once the first microcode fails to load, the second try only occurs a minute or so later. By that time, the bootup process is over, the network scripts are done running and have failed to connect to anything.

I don't know if that should be a bug filed against the driver. Maybe it could try alternate APIs faster.

Comment 5 Thomas Backlund 2016-10-18 22:03:08 CEST

Kernel slow loading due to missing firmwares are fixed in:
Should be fixed with kernel-4.4.22-1.mga5 released in:
http://advisories.mageia.org/MGASA-2016-0345.html

Status: NEW => RESOLVED
Resolution: (none) => FIXED