This is the mail archive of the binutils@sourceware.org mailing list for the binutils 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: 68hc11/12/s12x/xgate patch


To follow up on Fred's email, some of the results in the spreadsheet linked were not accurate. The reason is that when I tried to assemble my test code I got "Relocation 14 is not supported by object file format". This error was caused by having a typo in the label identifier. The error message is non-intuitive and made it difficult to notice the mistake, especially considering the volume of other errors. I apologize for the mistake, and misleading information given, however these errors still seem to stand:

xgate.s: Assembler messages:
xgate.s:329: Error: Opcode `bhs' is not recognized.
xgate.s:333: Error: Opcode `blo' is not recognized.
xgate.s:346: Error: Opcode `com' is not recognized.
xgate.s:347: Error: Opcode `cpc' is not recognized.
xgate.s:348: Error: Opcode `cpc' is not recognized.
xgate.s:362: Error: Garbage at end of instruction: `ErrorHandler'. In my opinion, 'ldw' should be added as an alias to 'ld' as per the CPU reference manual.
xgate.s:370: Error: Opcode `mov' is not recognized.
xgate.s:371: Error: Opcode `neg' is not recognized.
xgate.s:400: Error: Opcode `tst' is not recognized.


I have some technical questions to ask on this subject too, and will send those at a later date.

Regards,
Sean

On 12/21/2011 03:29 PM, Fred Cooke wrote:
On Tue, 20 Dec 2011 23:29:37 +0100 Fred Cooke wrote:
To the maintainers: I would recommend holding off a little longer on
accepting this into the upstream repository. I head a project that
uses these tools and I believe there are some issues with this patch
that are yet to be shown.
Could you be more specific? What leads you to believe there are issues
with this patch?
You missed a bit, James!

There should be more information available inside the next 24 hours. Hopefully that is not too long to wait.

Thank you in advance for your patience.
I take it you're not feeling particularly patient, nevertheless, good
things take time, and rushed things are rarely good enough, as you'll
soon see. Feel free to skip the two paragraphs which amount to a
political history lesson and go right for the comparison link.

Way back in 2008 James released his initial work on an XGATE Binutils
patch and published it on the internet. At the time I was hard at work
on my project which also uses an XGATE core and, naturally, I was
quite interested in this work. I grabbed a copy of that work at the
time, however I can't find it right now. Shortly afterward I noticed
that all of the files had been pulled from circulation and requested
that James email me his work so that we could collaborate in testing
and development with him. I was not the only one to request these
files during that period, however nobody had their request fulfilled.
Although he was perfectly within his rights to do what he did, it
never struck me as a particularly moral thing to do with a
foundational GNU package. Late in 2010 James re-released his work, and
because of his earlier behaviour I decided to cache it here:

http://tools.diyefi.org/james/s12x-tools-source-20101029.zip

In the mean time, and in the absence of James sharing his changes with
other members of the community, Sean Keys started work on a parallel
effort to produce a working tool-chain for the S12X platform including
an XGATE assembler and linker. Sean assures me that he did extensive
manual checking and comparative checking to ensure that his work was
of the best possible quality. He says that he used CodeWarrior as a
bench mark as that is the suite that the manufacturer of the chipset
provides - they should know! When we saw that James was trying to
upstream his work, we decided, for the benefit of all future users,
and because it was the right thing to do, to give it a once over and
make sure that it functioned as advertised. Sean has assembled some
results in a spreadsheet which I've published on our tools page, which
has been up, since October 16, 2010, for more than a year.

http://tools.diyefi.org/XGATE-AssemblerComparison.html

I'll have to leave it to Sean to explain what he found in his testing
as I have some other tasks to complete today, however I hope that the
maintainers of Binutils will observe the missing instructions which
render James' tools unable to build other existing code bases for this
chipset. From cross examining Sean and from looking into some of this
behaviour myself, I believe that this patch is not ready for prime
time. Sean should be available to follow up on this email in an hour
or two or three. Again, I hope your patience extends that far, James
excluded, his clearly does not.

Regards,

Fred.

On Wed, Dec 21, 2011 at 12:22 AM, James Murray<jsm@jsm-net.demon.co.uk> wrote:
On Tue, 20 Dec 2011 23:29:37 +0100 Fred Cooke wrote:
To the maintainers: I would recommend holding off a little longer on
accepting this into the upstream repository. I head a project that
uses these tools and I believe there are some issues with this patch
that are yet to be shown.
Could you be more specific? What leads you to believe there are issues
with this patch?

It is a cleanup of the patch posted here in March and the same as the
tools on http://www.msextra.com/tools that have been in use for a few
years now.

regards

James



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