Bug 5681

Summary:	8139cp has broken suspend/resume function
Product:	Drivers	Reporter:	Pierre Ossman (pierre-bugzilla)
Component:	Network	Assignee:	Francois Romieu (romieu)
Status:	CLOSED CODE_FIX
Severity:	high	CC:	chenliunju, romieu
Priority:	P2
Hardware:	i386
OS:	Linux
Kernel Version:	2.6.? - 2.6.15-rc3	Subsystem:
Regression:	---	Bisected commit-id:
Attachments:	8139cp: fix broken suspend/resume 8139cp: misc cleanups

Description Pierre Ossman 2005-11-30 00:53:10 UTC

Most recent kernel where this bug did not occur:

Probably never since the code seems to have been added in the current state.

Distribution:

Fedora Core Rawhide

Hardware Environment:

HP/Compaq nx7010

Problem Description:

The resume function for 8139cp is horribly broken. It has two major problems:

 * It assumes the interface is up and activates the hardware. If the interface
isn't up, then inserting the network cable will cause an interrupt which in turn
kills the interrupt. A reboot of the machine is necessary to get things back to
normal.

 * It doesn't properly reset the hardware. At minimum a ifconfig down/up is
needed to get things going. Sometimes the module needs to be unloaded and
reloaded again. The effect is that nothing can be sent/recieved. No errors occur
so it probably thinks everything is peachy.

Steps to reproduce:

1. Suspend to disk.
2. Resume.
3. Swear in frustration when you cannot reconnect to the network.


I have a patch in -mm that forces the interrupt handler to be loaded at all
times (to solve the first problem). It's a dirty solution but was all I could
come up with on my own.

Comment 1 Francois Romieu 2006-02-23 15:45:38 UTC

Created attachment 7462 [details]
8139cp: fix broken suspend/resume

Comment 2 Francois Romieu 2006-02-23 15:49:21 UTC

Created attachment 7463 [details]
8139cp: misc cleanups

Pierre, can you give these two patches a try to check that I have not messed
anything ?

Thank in advance.

-- 
Ueimor

Comment 3 Pierre Ossman 2006-02-27 01:53:32 UTC

Works like a charm :)

Comment 4 L.C. 2010-05-05 03:35:14 UTC

I see that there are two patches 23397bee and cc799034 for this bug. But it seems that I can't find these two patches in git. Do you know the reason? Thanks!