Bug 5791

Summary:	pcnet_cs off-by-one buffers after link-down/link-up?
Product:	[Retired] Red Hat Linux	Reporter:	paul
Component:	kernel	Assignee:	Michael K. Johnson <johnsonm>
Status:	CLOSED CURRENTRELEASE	QA Contact:
Severity:	medium	Docs Contact:
Priority:	medium
Version:	6.1
Target Milestone:	---
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2000-02-18 00:42:50 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description paul 1999-10-09 21:54:29 UTC

RH 6.1
Dell Inspiron 3500
D-Link DMF-560TX combo ethernet/modem pcmcia card

Begin with system functioning normally. Ping a host on the
same ethernet segment, see normal ping times (less than
1ms).

Disconnect twisted pair ethernet cable for a few seconds,
then reconnect. Ping times are now 1000 msec + a fraction
of a msec, i.e., exactly 1 second late. Other network
performance (e.g., nfs, etc.) becomes terrible.

execute "ifconfig down up". Ping times return to normal
and other network performance returns to normal.

This behaves like an off-by-one buffer pointer problem
triggered by link-down/link-up. Maybe a packet doesn't
get transmitted until another packet gets queued behind it.

Comment 1 Cristian Gafton 2000-01-04 22:26:59 UTC

Assigned to dledford

Comment 2 David Hinds 2000-02-18 00:42:59 UTC

I think this is fixed in the current PCMCIA driver package.  It was fixed in
3.1.4.