Re: [PATCH v3] perf jevents: Parse metrics during conversion

[John Garry]

On 01/12/2022 03:41, Ian Rogers wrote:
> Currently the 'MetricExpr' json value is passed from the json
> file to the pmu-events.c. This change introduces an expression
> tree that is parsed into. The parsing is done largely by using
> operator overloading and python's 'eval' function. Two advantages
> in doing this are:
>
> 1) Broken metrics fail at compile time rather than relying on
>     `perf test` to detect. `perf test` remains relevant for checking
>     event encoding and actual metric use.

Do we still require the code to "resolve metrics" in resolve_metric()? 
But I'm not sure it even ever had any users.

> 2) The conversion to a string from the tree can minimize the metric's
>     string size, for example, preferring 1e6 over 1000000, avoiding
>     multiplication by 1 and removing unnecessary whitespace. On x86
>     this reduces the string size by 3,050bytes (0.07%).

Out of curiosity, did you try the exponent change on its own (to see the 
impact on size)?

Nit:

Unrelated, really, I notice that sometimes we lose the parenthesis and 
sometimes never had them, like:

/* offset=11526 */ "\000\000metrics\000Ave [...] 0\000( 1000000000 * ( 
UNC_CHA
/* offset=11207 */ "\000\000metrics\000Ave [...] 0\0001e9 * (UNC_CHA_TOR

To me, it seems neater to have the expression contained within (a 
parenthesis) ever since we moved to this "big string". This seems to be 
a preexisting feature.

Thanks,
John


> In future changes it would be possible to programmatically
> generate the json expressions (a single line of text and so a
> pain to write manually) for an architecture using the expression
> tree. This could avoid copy-pasting metrics for all architecture
> variants.
>
> Signed-off-by: Ian Rogers<irogers@xxxxxxxxxx>