rtems-libbsd: M68K linkcmds Patches Galore

Kevin Polulak kpolulak at gmail.com
Mon Aug 20 06:44:35 UTC 2012


On 8/17/2012 4:13 AM, Sebastian Huber wrote:
> I don't like this format.  Each time you change the length of one 
> field in the columns you have to adjust other lines as well.  This 
> makes a review harder.

So code should remain cryptic and hard to follow because git-diff 
formats things strange sometimes? A review of the .diff output may be 
hard but a review of the actual code is even harder. That's Git's fault. 
I was just trying to clean things up for the next poor guy who has to 
look at it.

Say "hello" to my logic:

I perform task A and I like to do task B and I do task C.
I do the same thing as A and I do something like B and I'm identical to C.

Versus:

I perform task A          and I like to do task B    and I sometimes do 
task C.
I do the same thing as A  and I do something like B  and I'm identical to C.

The tabular flow forms these invisible columns that 1) are easier on the 
eyes and 2) group similar things together.

+-----------------------------------------------------------------------------------+
|          TASK A          |           TASK B |           TASK C           |
+-----------------------------------------------------------------------------------+
| I perform task A         | and I like to do task B   | and I sometimes 
do task C. |
| I do the same thing as A | and I do something like B | and I'm 
identical to C.    |
+-----------------------------------------------------------------------------------+

(Hoping my email client doesn't screw up that formatting for me.)

> If you look at the default linker command file (e.g. m68k-rtems4.11-ld 
> --verbose) then you will see that the previous format is used here 
> (space before "(").

That's fine. I always say: I could care less what the format is, just so 
long as it's followed. Sometimes there was a space, sometimes there 
wasn't and it just seemed like there were more "no spaces" so that's 
what I went with. Easy change.

> Why is this linkcmds.robsdsets and not linkcmds.rortems?

Didn't even realize I did that. The name got stuck in my head from 
typing it so many times and I guess it stuck.

> Why do we have CPU specific variants of this?  On ARM, PowerPC, etc. 
> this file would look like identical.

You mean because it's in m68k/shared? I didn't know where else to put 
linkcmds.rortems. Joel said to use your's or Gedare's way (I have no 
clue, everyone's e-face blends together for me) from ARM so that's what 
I did. There's still about 95% of unexplored RTEMS territory for me so I 
didn't want to waste time trying to do something I didn't know how to. 
How do I make it "attention all BSP's, you use this linkcmds file and 
you use it now!"? ;)

And skipping forward to all this other GCC __underscore__ talk...

Hold on. Timeout.

To be honest, I don't even have a clue what that stuff is inside 
linkcmds.robsdsets. I did a git-grep on some of those symbols a while 
back and didn't find them referenced anywhere so their purpose has 
always eluded me. But they fix the linkage issues so that's just what I 
put in there.

But I'm not sure how the conversation got on this though? I mean, 
whatever that stuff is, it works. What I was struggling with is 
including it in an external file (i.e. linkcmds.rortems). So somehow 
that train of thought didn't make a stop at my boarding station and I'm 
kinda still waiting to be picked up. :/

(By the way, this 
<http://sourceware.org/binutils/docs-2.17/ld/Source-Code-Reference.html#Source-Code-Reference> 
says it's a compiler "feature".)

-- 
- Kevin Polulak (soh_cah_toa)
- http://cybercrud.net

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/devel/attachments/20120820/c479bb42/attachment-0001.html>


More information about the devel mailing list