MIME-Version: 1.0
References: <20230204134651.22569-1-namcaov@gmail.com> <Y95jtuMbcDiPRw/F@kroah.com>
In-Reply-To: <Y95jtuMbcDiPRw/F@kroah.com>
From:   Nam Cao <namcaov@gmail.com>
Date:   Sat, 4 Feb 2023 15:01:14 +0100
Message-ID: <CA+sZ8B9Wrr4yj3c6fiqFhKvZuYifV87EXBQ1pY_Km-p=3oE4ug@mail.gmail.com>
Subject: Re: [PATCH] USB: serial: whiteheat: use stack instead of heap memory
To:     Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc:     Johan Hovold <johan@kernel.org>, linux-usb@vger.kernel.org,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Sat, Feb 4, 2023 at 2:55 PM Greg Kroah-Hartman
<gregkh@linuxfoundation.org> wrote:
>
> On Sat, Feb 04, 2023 at 02:46:51PM +0100, Nam Cao wrote:
> > Some buffers in whiteheat_attach() are small and only used locally. Move
> > them to the stack to avoid complications with heap memory.
> >
> > Compile-tested only.
>
> And that's the problem, you can't just compile test these things, the
> code will blow up if you make these changes :(
>
> All USB transfers need to come from memory that can be safely DMAed.
> Stack memory is not that type of memory, you HAVE to allocate it
> dynamically from the heap in order to have this guarantee.
>
> So no, this patch is not acceptable, sorry.  You will see this pattern
> in all USB drivers, all data must be dynamically allocated, even for 2
> byte commands.
>
> So yes, there was a reason we added this "complexity" to the driver, it
> is required :)

Thanks for "the lecture", and sorry for the broken patch.

Best regards,
Nam