This is the mail archive of the mailing list for the eCos project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: AW: another wakep call

Hi Norbert,

You're probably not going to like this, but, wow, you're 
making it even harder to review by sending everything. I 
suppose you mean anoncvs of august 14th, with 14.8.05 trunk? 
If this is the case, just cd to the packages directory and 

cvs diff -U -N5 > my.patch

on the command line. The file my.patch now only contains what 
you modified (search this list, you'll find numerous examples 
of such patch files). Pls send patches in this format to this 
list, otherwhise, chances are still odd that anyone will ever 
look at them, simply because it takes too much time finding 
out what exactly the modifications are. 

As an alternative, you can place a file called .cvsrc in your 
home dir and add a line

diff -U -N5

to it. All you have to do now is enter cvs diff > my.patch on 
the command line. Also don't forget to make a ChangeLog entry 
when fixing things.

Thanks for reporting problems.


---- Original message ----
>Date: Wed, 17 Aug 2005 09:48:47 +0200
>From: "Norbert Peschel" <>  
>Subject: AW: another wakep call  
>To: <>
>Hi Gary, hi Bob
>Plus, it is much effort to patch each version again ;)
>You can compare/replace the code directly to 14.8.05 trunk.
>kernel.h      : brackets missing in macros
>tftp_server.c : we have ram addresses starting at 0x80000000
>                ehich makes success checks for < 0 not 
>httpd.c	  : good-weather-code, especially getc() can 
>                return -1 anytime which will loop endlessly
>                and block all lower priority tasks (e.g. ppp)
>pppd.c	  : I ended up seeing your linked timeout-list
>                screwed up (circular), this all vanished with
>                the mutex around its access
>		    I found an echo-Reply timeout event left 
on the 
>                list in case the line timed out, if that 
>		    too often the line immediately times out 
on re-
>		    connect, fixed it by cleaning up the list 
>                manually (cleantimeout)
>sys-ecos      : I saw the pppd_main immediatly failing (due 
>                the errors above) and ppp_up() waiting 
>                für !PHASE_DEAD long after pppd_main had 
>The ppp-fixes are certainly not what you want (fixing effects
>of bugs) but for me they to their job.
>Regards, Norbert
>-----Ursprüngliche Nachricht-----
>Von: Gary Thomas [] 
>Gesendet: Mittwoch, 17. August 2005 00:47
>Cc: Norbert Peschel; eCos patches
>Betreff: Re: another wakep call
>On Tue, 2005-08-16 at 07:40 -0400, 
>> Why don't you not propose a solution and send a patch? 
>> (that's what the list is called anyway, ecos-patches :-) 
Makes it much 
>> easier/less time consuming for the maintainers to review 
and apply if 
>> considered appropriate.
>Plus, you've (Norbert) analyzed the failure and know at 
least one way to solve it.  The maintainers can then look at 
this and decide whether to accept your patch directly (likely 
if it's
>reasonable) or use it as information to develop a different 
>As Bob said, that's what this list (and eCos) is all about.
>We've shared the sources with the world, it's only fair when 
the world share their improvements, etc.
>Gary Thomas                 |  Consulting for the
>MLB Associates              |    Embedded world
> 51k bytes

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]