qemu-devel
[Top][All Lists]
Advanced

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

Re: [PATCH v2] decodetree: Allow use of hex/bin format for argument fiel


From: Richard Henderson
Subject: Re: [PATCH v2] decodetree: Allow use of hex/bin format for argument field values
Date: Tue, 1 Dec 2020 06:31:19 -0600
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.10.0

On 12/1/20 4:34 AM, Philippe Mathieu-Daudé wrote:
> ISA datasheets often use binary or hexadecimal constant values.
> By doing base conversion, we might introduce bugs. Safer is to
> copy/paste the datasheet value.
> 
> To add support for bin/hex constants in argument field token,
> extend the re_num_ident regexp and use '0' as radix base when
> parsing the integer (to interpret it as a code literal).
> 
> Suggested-by: Richard Henderson <richard.henderson@linaro.org>
> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
> ---
> v2: Fix regexp (Richard)
> ---
>  scripts/decodetree.py | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/scripts/decodetree.py b/scripts/decodetree.py
> index 47aa9caf6d1..c8136073535 100644
> --- a/scripts/decodetree.py
> +++ b/scripts/decodetree.py
> @@ -50,6 +50,7 @@
>  re_fld_ident = '%[a-zA-Z0-9_]*'
>  re_fmt_ident = '@[a-zA-Z0-9_]*'
>  re_pat_ident = '[a-zA-Z0-9_]*'
> +re_num_ident = '[+-]?([0-9]+|0x[0-9a-fA-F]+|0b[01]+)'

re_num or re_number, perhaps -- it's not an identifier.

>          # 'Foo=number' sets an argument field to a constant value
> -        if re.fullmatch(re_C_ident + '=[+-]?[0-9]+', t):
> +        if re.fullmatch(re_C_ident + '=' + re_num_ident, t):
>              (fname, value) = t.split('=')
> -            value = int(value)
> +            value = int(value, 0)
>              flds = add_field(lineno, flds, fname, ConstField(value))
>              continue

It occurs to me that we don't actually have to interpret this string --
decodetree doesn't do anything with it but pass it along to the generated C
code.  We could in fact accept a re_C_ident as well, to allow symbolic values.
 So long as the symbols are defined before the decode.c.inc file is included.

So perhaps just change ConstField to not expect an integral value, and just
store the string.


r~



reply via email to

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