bug-binutils
[Top][All Lists]
Advanced

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

Fwd: Lost and Wondering


From: Matthew Carpenter
Subject: Fwd: Lost and Wondering
Date: Fri, 18 Jun 2010 16:16:16 -0400
User-agent: KMail/1.13.2 (Linux/2.6.32-22-generic; KDE/4.4.2; x86_64; ; )

Is this the right place for this?
--- Begin Message --- Subject: Lost and Wondering Date: Fri, 18 Jun 2010 12:20:24 -0400 User-agent: KMail/1.13.2 (Linux/2.6.32-22-generic; KDE/4.4.2; x86_64; ; )
I'm confused about a problem I'm running into.

I have been working on tools to disassemble ARM instructions and am running 
into a discrepancy between the ARM documentation and the GNU assembler, 
particularly over the ldr..sh and ldr..sb instructions.  The GNU assembler 
takes the following instruction (for example):

        ldreqsh r9, [sl], #0x3f

and assembles it to:

        0x00da94d0

when the documentation plainly requires the penultimate nibble to be 0xf and 
the third nibble (also a 0xd here) would indicate that the immediate offset 
should be subtracted (U bit is set).
With the penultimate nibble being 0xd, this instruction (according to the ARM 
documentation) should be ldreqsb.

I subsequently was testing the ldreqsb instruction (this is an automated tool 
so the operands are slightly different), and found that the following 
instruction:
        
        ldreqsb r9, [sl], #0x40

assembled as:
        
        0x00da93ff

which, accordingly should correspond to the ldreqsh instruction.  

I don't have an environment set up currently to play with to prove empirically 
which is correct, so I wanted to bring it to you both, the experts.  Ordinarily 
I would only approach Mr. Raeburn, however other disassembler tools break down 
the bytes in the same way as arm-elf-as assembles them, and I have found flaws 
in other documentation for other <ahem> bigger microcontrollers whose Intel 
igence shall go unnamed.  

Through my automated testing, I have found this bug to generate over half of my 
error messages (and it's the error messages that count in this tool).

Please help?  Which is accurate?  Should I trust the assembler, which is 
responsible for the actual driving of the processor?  or the documentation 
which is the Bible of ARM?

Thank you,
Matt

-- 
Matthew Carpenter
address@hidden
http://www.inguardians.com

PGP Fingerprint: 
87EB 54A8 FB42 0A0E B8AE CDA7 FF99 2A64 E70F 4466
hkp://wwwkeys.pgp.net

--- End Message ---

reply via email to

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