On Sat, Feb 20, 2021 at 8:33 AM Ben Widawsky <ben.widawsky(a)intel.com> wrote:
On 21-02-19 20:22:00, Konrad Rzeszutek Wilk wrote:
> ..snip..
> > +static int handle_mailbox_cmd_from_user(struct cxl_mem *cxlm,
> > + const struct cxl_mem_command *cmd,
> > + u64 in_payload, u64 out_payload,
> > + s32 *size_out, u32 *retval)
> > +{
> > + struct device *dev = &cxlm->pdev->dev;
> > + struct mbox_cmd mbox_cmd = {
> > + .opcode = cmd->opcode,
> > + .size_in = cmd->info.size_in,
> > + .size_out = cmd->info.size_out,
> > + };
> > + int rc;
> > +
> > + if (cmd->info.size_out) {
> > + mbox_cmd.payload_out = kvzalloc(cmd->info.size_out,
GFP_KERNEL);
> > + if (!mbox_cmd.payload_out)
> > + return -ENOMEM;
> > + }
> > +
> > + if (cmd->info.size_in) {
> > + mbox_cmd.payload_in = vmemdup_user(u64_to_user_ptr(in_payload),
> > + cmd->info.size_in);
> > + if (IS_ERR(mbox_cmd.payload_in))
> > + return PTR_ERR(mbox_cmd.payload_in);
>
> Not that this should happen, but what if info.size_out was set? Should
> you also free mbox_cmd.payload_out?
>
Thanks Konrad.
Dan, do you want me to send a fixup patch? This bug was introduced from v4->v5.
Yes, please, incremental to libnvdimm-for-next which I'm planning to
send to Linus on Tuesday.