This is the mail archive of the
ecos-patches@sourceware.org
mailing list for the eCos project.
another wakep call
- From: "Norbert Peschel" <N dot Peschel at conergy dot de>
- To: <ecos-patches at sources dot redhat dot com>
- Date: Tue, 16 Aug 2005 12:28:18 +0200
- Subject: another wakep call
Hi there
It seems not really much work to forward a bug-report email.
But obviously more than you want to invest (see email below).
I will not spend my time to report in detail if noone takes
interest of errors which are plain to see and which take no
time to correct. If you want to know more details just email
back.
Regards, N. Peschel
ppp-stack:
is buggy, access to the linked timeout list *does* need a mutex to
prevent circular lists and such. whenever you have a echo reply
timeout you leave a timeout on that timeoutlist. So if you have
"collected" enough old timeouts the dialin will be shutdown imme-
diately after a new dialin. I solved the problem flushing that
list before each dialin. That works fine with about 5000 checked
dialins.
httpd:
uses a return value of a function which can return -1 as an index
on top level.
-----Ursprüngliche Nachricht-----
Von: Norbert Peschel
Gesendet: Mittwoch, 6. Juli 2005 16:04
An: 'info@ecoscentric.com'
Betreff: patch hints
Hi, I guess I found a few issues:
***
pppd.c:
if (newp->c_time.tv_sec < p->c_time.tv_sec
|| (newp->c_time.tv_sec == p->c_time.tv_sec
&& newp->c_time.tv_usec < p->c_time.tv_sec))
I assume you rather wanted compare usec with usec in the last line?!
***
kernel.h
#define min(x,y) (x<y?x:y)
#define max(x,y) (x<y?y:x)
lacking brackets around the arguments these macros result potentially in wrong code.
***
Hope this helps a bit, great software
kind regards, Norbert Peschel
--------------------------------------------------------
Norbert Peschel
Firmwareentwicklung
Conergy AG
Anckelmannsplatz 1
D 20537 Hamburg
Tel +49 40 237102 198
Fax +49 40 237 102 148
www.conergy.de
--------------------------------------------------------