gnats-prs
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

gnats/259: Text added to "Reason Changed" and "Fix" fields disappears af


From: dave . kissel
Subject: gnats/259: Text added to "Reason Changed" and "Fix" fields disappears after edit submitted
Date: 14 Sep 2001 16:51:57 -0000

>Number:         259
>Category:       gnats
>Synopsis:       Text added to "Reason Changed" and "Fix" fields disappears 
>after edit submitted
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Fri Sep 14 09:54:00 PDT 2001
>Closed-Date:
>Last-Modified:
>Originator:     Dave Kissel
>Release:        unknown-1.0
>Organization:
>Environment:

>Description:
In an attempt to change the state of PR #2385 for the PMAC project, the PR was 
opened for editing. The state was changed to "feedback." The text "Root cause 
found" was added to the "Reason Changed" field under the state. The text 
following this text was added to the "Fix" field. The edit was then submitted. 
Gnats reported "edit successful" and the page was refreshed. The PR was then 
viewed to discover that the state change successfully changed to "feecback," 
but that the "Reason Changed" and "Fix" fields were blank. The information was 
lost.

As the "fix" description was lengthy, a copy of it was saved after entering it. 
Several attempts to re-edit and paste the fix info into the fix field also 
failed. The author waited overnight to see if there was a time delay in the 
information appearing. It never did.

Other developers were able to add info to the fix field. When the author 
subsequently attempted to re-add his fix info, only the info added by the other 
developer showed-up after the ("successful") edit session.

The text the author attempted to add was:
The failure to recover from loopback plug removal is due to the "lockup" of the 
MAC device. The MAC is programmed with two transmit threshold values by the SA 
Core. These are the "Transmit Threshold (TSD)" in the "Transmit Threshold and 
Backoff Register" and the "Transmit FIFO Threshold (TTH)" in the "FIFO 
Threshold Register." Packet transmission starts when the number of bytes in the 
transmit FIFO crosses the TSD threshold. The transmit ready flag is asserted 
when the number of bytes in the transmit FIFO drops below the TTH threshold. 
IXF440 documentation for the Transmit Threshold and Backoff Register states 
that "If the sum of the effective serial transmit threshold and the effective 
IX Bus transmit threshold (based upon FFO_TSHD<TTH>) is larger than 230 bytes, 
transmission can begin before the effective serial transmit threshold is 
reached in order to prevent deadlock." The values originally written to the TTH 
and TSD fields did not meet this condition. Thus, it is presumed that pulling 
the loopback plug caused the transmit FIFO to get into a state where transmit 
ready was not asserted, preventing the uEngines from writing more bytes into 
the transmit FIFO, but the number of bytes in the transmit FIFO had not yet 
reached the TSD threshold. This caused the MAC to "deadlock." Increasing the 
TTH field from 7 to 9 met this condition and eliminated the lockup that occured 
when the loopback plug was removed and reinserted.

>How-To-Repeat:

>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted:


reply via email to

[Prev in Thread] Current Thread [Next in Thread]