On Wed, Mar 25, 2026 at 07:36:27PM -0700, Rosen Penev wrote:
> Use a flexible array member to combine allocations into one.
>
> Add __counted_by for extra runtime analysis. Move counting variable
> assignments to after allocation as required by __counted_by.
I don't think it improves readability of the code. Also see below.
...
> - vg = devm_kzalloc(dev, sizeof(*vg), GFP_KERNEL);
> + vg = devm_kzalloc(dev, struct_size(vg, communities,
> soc_data->ncommunities), GFP_KERNEL);
> if (!vg)
> return -ENOMEM;
>
> + vg->ncommunities = soc_data->ncommunities;
> + memcpy(vg->communities, soc_data->communities, soc_data->ncommunities *
> sizeof(*vg->communities));
This is a mess. The multiplication is used without overflow check and the parts
of it are from different data structures. If we do a such, we need to use
source data for the all of the information.
What I would like to see is rater special devm_kmalloc_and_dup()-like API
instead of doing this in every driver out of dozens of drivers.
...
> - ret = byt_set_soc_data(vg, soc_data);
> - if (ret)
> - return dev_err_probe(dev, ret, "failed to set soc data\n");
> + vg->soc = soc_data;
> +
> + for (i = 0; i < vg->soc->ncommunities; i++) {
> + struct intel_community *comm = vg->communities + i;
> +
> + comm->pad_regs = devm_platform_ioremap_resource(pdev, 0);
> + if (IS_ERR(comm->pad_regs))
> + return PTR_ERR(comm->pad_regs);
> + }
Why is this can't be kept in a separate function as it was originally done?
--
With Best Regards,
Andy Shevchenko
