Re: [PATCH 4.19 111/264] nvmem: core: fix possibly memleak when use nvmem_cell_info_to_nvmem_cell()

From: Pavel Machek
Date: Wed Oct 28 2020 - 18:42:00 EST


Hi!

> From: Vadym Kochan <vadym.kochan@xxxxxxxxxxx>
>
> [ Upstream commit fc9eec4d643597cf4cb2fef17d48110e677610da ]
>
> Fix missing 'kfree_const(cell->name)' when call to
> nvmem_cell_info_to_nvmem_cell() in several places:
>
> * after nvmem_cell_info_to_nvmem_cell() failed during
> nvmem_add_cells()
>
> * during nvmem_device_cell_{read,write} when cell->name is
> kstrdup'ed() without calling kfree_const() at the end, but
> really there is no reason to do that 'dup, because the cell
> instance is allocated on the stack for some short period to be
> read/write without exposing it to the caller.
>
> So the new nvmem_cell_info_to_nvmem_cell_nodup() helper is introduced
> which is used to convert cell_info -> cell without name duplication as
> a lighweight version of nvmem_cell_info_to_nvmem_cell().
>
> Fixes: e2a5402ec7c6 ("nvmem: Add nvmem_device based consumer apis.")

There's something very wrong here.

> index 30c040786fde2..54204d550fc22 100644
> --- a/drivers/nvmem/core.c
> +++ b/drivers/nvmem/core.c
> @@ -326,9 +326,9 @@ static void nvmem_cell_add(struct nvmem_cell *cell)
> mutex_unlock(&nvmem_cells_mutex);
> }
>
> -static int nvmem_cell_info_to_nvmem_cell(struct nvmem_device *nvmem,
> - const struct nvmem_cell_info *info,
> - struct nvmem_cell *cell)
> +static int nvmem_cell_info_to_nvmem_cell_nodup(struct nvmem_device *nvmem,
> + const struct nvmem_cell_info *info,
> + struct nvmem_cell *cell)
> {
> cell->nvmem = nvmem;
> cell->offset = info->offset;
> @@ -345,13 +345,30 @@ static int nvmem_cell_info_to_nvmem_cell(struct nvmem_device *nvmem,
> if (!IS_ALIGNED(cell->offset, nvmem->stride)) {
> dev_err(&nvmem->dev,
> "cell %s unaligned to nvmem stride %d\n",
> - cell->name, nvmem->stride);
> + cell->name ?: "<unknown>", nvmem->stride);
> return -EINVAL;
> }
>
> return 0;
> }

We rename call from .._cell to .._cell_nodup, but it did not have the
kstrdup_const() in the first place!

> +static int nvmem_cell_info_to_nvmem_cell(struct nvmem_device *nvmem,
> + const struct nvmem_cell_info *info,
> + struct nvmem_cell *cell)
> +{
> + int err;
> +
> + err = nvmem_cell_info_to_nvmem_cell_nodup(nvmem, info, cell);
> + if (err)
> + return err;
> +
> + cell->name = kstrdup_const(info->name, GFP_KERNEL);
> + if (!cell->name)
> + return -ENOMEM;
> +
> + return 0;
> +}

So now we introduce an allocation, but we don't have a place to free
it. In mainline, it is freed in nvmem_cell_drop(), but 4.19 does not
have a free there.

Best regards,
Pavel
--
http://www.livejournal.com/~pavelmachek

Attachment: signature.asc
Description: PGP signature