netdev - Re: [PATCH net] gve: DQO: Suppress unused var warnings

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAK8P3a1P9ZGBCub2s62OjrUa1Hwk66zHHESEj06MPY8qwjK7Ag@mail.gmail.com>
Date:   Tue, 28 Sep 2021 16:15:06 +0200
From:   Arnd Bergmann <arnd@...nel.org>
To:     Bailey Forrest <bcf@...gle.com>
Cc:     Jakub Kicinski <kuba@...nel.org>,
        "David S . Miller" <davem@...emloft.net>,
        Networking <netdev@...r.kernel.org>
Subject: Re: [PATCH net] gve: DQO: Suppress unused var warnings

On Tue, Sep 28, 2021 at 2:00 AM Bailey Forrest <bcf@...gle.com> wrote:
> On Mon, Sep 27, 2021 at 4:21 PM Jakub Kicinski <kuba@...nel.org> wrote:
> > On Mon, 27 Sep 2021 13:21:30 -0700 Bailey Forrest wrote:
> >
> > Looks like fixing this on the wrong end, dma_unmap_len_set()
> > and friends should always evaluate their arguments.
>
> That makes sense.
>
> Arnd, if you want to fix this inside of the dma_* macros, the
> following diff resolves the errors reported for this driver:

> diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h
> index dca2b1355bb1..f51eee0f678e 100644
> --- a/include/linux/dma-mapping.h
> +++ b/include/linux/dma-mapping.h
> @@ -590,10 +590,21 @@ static inline int dma_mmap_wc(struct device *dev,
>  #else
>  #define DEFINE_DMA_UNMAP_ADDR(ADDR_NAME)
>  #define DEFINE_DMA_UNMAP_LEN(LEN_NAME)
> -#define dma_unmap_addr(PTR, ADDR_NAME)           (0)
> -#define dma_unmap_addr_set(PTR, ADDR_NAME, VAL)  do { } while (0)
> -#define dma_unmap_len(PTR, LEN_NAME)             (0)
> -#define dma_unmap_len_set(PTR, LEN_NAME, VAL)    do { } while (0)
> +
> +#define dma_unmap_addr(PTR, ADDR_NAME) ({ (void)PTR; 0; })
> +
> +#define dma_unmap_addr_set(PTR, ADDR_NAME, VAL) do { \

Unfortunately, this breaks every other driver using these macros, as the
point of them is that the unmap-address is not actually defined
and not taking up space in data structure. Referencing it by name
is a compile-time bug.

I've come up with a new patch to gve that just removes the
"struct gve_tx_dma_buf" and open-codes the access everywhere,
sending that as v2 now. Feel free to take that and modify as needed
before sending on, or doing yet another patch.

        Arnd