Subject:
Re: gpasm bugs
From:
Craig Franklin ####@####.####
Date:
26 Mar 2002 02:38:25 -0000
Message-Id: <3C9FDDEC.EEC77530@attbi.com>
Jesse Lackey wrote:
>
> Also, thanks to everyone who has spent time and effort to create and
> support these tools. I'm delighted that I can do effective PIC work in
> linux.
>
> Jesse
>
Thanks for the message. The feedback really helps.
The gputils sourceforge site doesn't have any open bug reports, but that
doesn't prevent you from opening a new report.
> gpasm suggestions, for gpasm -version = "0.10.1 alpha"
>
> "movf RamLocation1, Ramlocation2" is allowed but not legal. gpasm evals
> the RHS, and if 0 goes to w or if 1 to 'f'. I suggest having the RHS be
> required to be 0,1,w,f, only. Does this make sense? I would think it
> very unusual to have some expression on the RHS determine if the
> destination is w or f.
I don't believe mpasm limits the RHS to 0, 1, w, or f, so gpasm won't.
I will check this.
If the symbol "Ramlocation2" is a value other than 0 or 1 gpasm should
generate a warning, but it doesn't. It will, however, generate the same
"wrong" opcode that mpasm does. This is an easy feature to add. I will
take care of it in 0.10.2. Give me a week or two.
>
> The trick of using "^" to produce warnings if a register in bank 1 is
> possibly being accessed as though it is in bank 0 doesn't seem to work.
> In the line of code below, if the "^ 0x80" isn't there, MPASM will
> produce the message "Register in operand not in bank 0. Ensure that
> bank bits are correct". This is a nice feature, and seems to be in
> common use. (see page 71 of Predko's programming and customizing PIC
> microcontrollers book)
>
> btfss TXSTA ^ 0x80, TRMT
>
> Am I understanding this correctly? gpasm issues no warnings and
> produces the same output for "btfss TXSTA^0x80, TRMT" and "btfss TXSTA,
> TRMT".
gpasm doesn't generate any bank warnings. That is why your example
isn't working. I will look into this for 0.10.2.
>
> gpasm -help : for -w, -a, and -r settings, say what the default is.
OK.
> Also what the message levels mean, does level 0 mean more or less
> messages than 2? (you have to find / eventually stumble across the
> ERRORLEVEL directive documentation to find this out).
>
The message levels provide a global way to suppress messages and
warnings. It works that same way that mpasm's message levels do. 0 (the
default) reports everything, 1 reports warnings and errrors, and 2
reports only the errors.
If a gpasm feature explaination seems lacking, read mpasm users manual.
> gpasm documentation (I'm reading the .pdf):
> (1)
> Section 1.1, options in table don't have a "-" preceding them, except
> for setting the warning level. The "-w" should be "w" (or each option
> should be shown with a hyphen). Ya okay this is extremely picky of me.
"-w" should be "w".
>
> (2)
> Section 1.2.3, how to define the radix, gpasm also accepts the MPASM
> style "30h" for 30 hexadecimal, this is not mentioned. And "30d" for 30
> decimal presumably?
mpasm has many features which aren't documented. I believe that this is
one of them. I have rev g of their manual and I don't see it in there.
gpasm implements many of these features, but very few are in the
manual.
I get the impression that the mpasm users manual only documents the
recommended features. Even though mpasm still supports the legacy
features. This specific feature has the potential to cause some
confusion. Is 3d equal to 3 decimal or 3d hex? I believe mpasm treats
it as 3d hex if the radix is hex, but 3 decimal for all other radices.
gpasm does the same. Without knowledge of how the assembler works, the
source code is ambiguous.
This doesn't prevent the feature from being documented. I will try to
find some time to do it.
>
> (3)
> Say what the defaults are for LIST directive.
OK.
>
> --
> Bill Gates is a crook and millions of tax dollars have been spent
> prosecuting and convicting Microsoft of illegal business practices. Do
> not give them another dime until they have repaid the US government
> these costs.
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ####@####.####
> For additional commands, e-mail: ####@####.####
