Re: [RFC PATCH] scripts/tracetool: don't barf validating TCG types

[Stefan Hajnoczi <stefanha@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 1632 bytes --]

On Tue, Apr 06, 2021 at 05:53:07PM +0100, Alex Bennée wrote:
> TCG types will be transformed into the appropriate host types later on
> in the tool. Try and work around this by detecting those cases and
> pressing on.
> 
> [AJB: this seems a bit too hacky - but the problem is validate_type is
> buried a few layers deep. Maybe we should just drop TCGv from
> ALLOWED_TYPES and manually do if bit.startswith("TCGv_") in validate_type?]

Please include a line from a trace-events file that triggers the issue.
It's unclear to me what the problem is although I guess it's related to
TCGv_* types being rejected by validate_type. The
bit.startswith("TCGv_") change you mentioned sounds fine or a slightly
more general ALLOWED_TYPE_PREFIXES = ['TCGv_'].

> Fixes: 73ff061032 ("trace: only permit standard C types and fixed size integer types")
> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
> Cc: Matheus Ferst <matheus.ferst@eldorado.org.br>
> ---
>  scripts/tracetool/__init__.py | 13 +++++++++----
>  1 file changed, 9 insertions(+), 4 deletions(-)
> 
> diff --git a/scripts/tracetool/__init__.py b/scripts/tracetool/__init__.py
> index 5bc94d95cf..ea078e34b4 100644
> --- a/scripts/tracetool/__init__.py
> +++ b/scripts/tracetool/__init__.py
> @@ -87,10 +87,11 @@ def out(*lines, **kwargs):
>      "ssize_t",
>      "uintptr_t",
>      "ptrdiff_t",
> -    # Magic substitution is done by tracetool
> +    # Magic substitution is done by tracetool TCG_2_HOST

This makes it clearer that "TCG_2_HOST" is a code reference:

  # Magic substitution is done using tracetool.transform.TCG_2_HOST

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]