[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH] decodetree: Allow 'dot' in opcode names
From: |
Philippe Mathieu-Daudé |
Subject: |
Re: [PATCH] decodetree: Allow 'dot' in opcode names |
Date: |
Tue, 12 Jan 2021 23:19:49 +0100 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.6.0 |
On 1/12/21 10:05 PM, Eduardo Habkost wrote:
> On Tue, Jan 12, 2021 at 07:41:56PM +0100, Philippe Mathieu-Daudé wrote:
>> Some ISA use a dot in their opcodes. Allow the decodetree
>> script to process them. The dot is replaced by an underscore
>> in the generated code.
>
> Will something break if we just use underscores instead of dots
> in the input file?
No, but then the opcode doesn't really match the spec.
>
>>
>> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
>> ---
>> scripts/decodetree.py | 3 ++-
>> 1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/scripts/decodetree.py b/scripts/decodetree.py
>> index 47aa9caf6d1..b7572589e64 100644
>> --- a/scripts/decodetree.py
>> +++ b/scripts/decodetree.py
>> @@ -49,7 +49,7 @@
>> re_arg_ident = '&[a-zA-Z0-9_]*'
>> re_fld_ident = '%[a-zA-Z0-9_]*'
>> re_fmt_ident = '@[a-zA-Z0-9_]*'
>> -re_pat_ident = '[a-zA-Z0-9_]*'
>> +re_pat_ident = '[a-zA-Z0-9_.]*'
>
> If pattern identifiers are going to follow different rules,
> doesn't this need to be documented at docs/devel/decodetree.rst?
I checked and luckily for me the opcode pattern identifiers is
not documented <:)
>
>>
>> def error_with_file(file, lineno, *args):
>> """Print an error message from file:line and args and exit."""
>> @@ -1082,6 +1082,7 @@ def parse_file(f, parent_pat):
>> elif re.fullmatch(re_fmt_ident, name):
>> parse_generic(start_lineno, None, name[1:], toks)
>> elif re.fullmatch(re_pat_ident, name):
>> + name = name.replace('.', '_')
>> parse_generic(start_lineno, parent_pat, name, toks)
>
> Do we want error messages generated by the script to use the
> modified identifier with underscores, or the original identifier
> with dots? (This patch does the former)
You are right, we want the former in the error message (the input
format).
Thanks,
Phil.