linux-kernel - Re: [PATCH v2] mmc: cavium: Improve request handling by proper use of API

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] [day] [month] [year] [list]

Date:   Tue, 23 Nov 2021 15:00:04 +0100
From:   Ulf Hansson <ulf.hansson@...aro.org>
To:     Wojciech Bartczak <wbartczak@...vell.com>
Cc:     linux-mmc@...r.kernel.org, rric@...nel.org, beanhuo@...ron.com,
        tanxiaofei@...wei.com, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] mmc: cavium: Improve request handling by proper use of API

On Mon, 15 Nov 2021 at 22:54, Wojciech Bartczak <wbartczak@...vell.com> wrote:
>
> The driver for cavium/marvell platforms uses directly mrq->done() callback
> to signalize the request completion. This method to finalize request
> processing is not correct.
>
> Following fix introduces proper use of mmc_request_done() API for
> all paths involved into handling MMC core requests.
>
> Changes v1 => v2:
> - Added missing variable slot and functionality to retrive
>   slot base on bus_id contained in response status register.

Version history is great, but should come outside the actual commit
message. See below.

>
> Signed-off-by: Wojciech Bartczak <wbartczak@...vell.com>
> ---

Version history should come here along with other information. Then
add the below three dashes to end the section.

---

>  drivers/mmc/host/cavium.c | 9 ++++++---
>  1 file changed, 6 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/mmc/host/cavium.c b/drivers/mmc/host/cavium.c
> index 95a41983c6c0..674cfaf5d64e 100644
> --- a/drivers/mmc/host/cavium.c
> +++ b/drivers/mmc/host/cavium.c
> @@ -435,8 +435,10 @@ static void cleanup_dma(struct cvm_mmc_host *host, u64 rsp_sts)
>  irqreturn_t cvm_mmc_interrupt(int irq, void *dev_id)
>  {
>         struct cvm_mmc_host *host = dev_id;
> +       struct cvm_mmc_slot *slot;
>         struct mmc_request *req;
>         u64 emm_int, rsp_sts;
> +       int bus_id;
>         bool host_done;
>
>         if (host->need_irq_handler_lock)
> @@ -456,6 +458,8 @@ irqreturn_t cvm_mmc_interrupt(int irq, void *dev_id)
>                 goto out;
>
>         rsp_sts = readq(host->base + MIO_EMM_RSP_STS(host));
> +       bus_id = get_bus_id(rsp_sts);
> +       slot = host->slot[bus_id];  /* bus_id is in a range 0..2 */
>         /*
>          * dma_val set means DMA is still in progress. Don't touch
>          * the request and wait for the interrupt indicating that
> @@ -493,8 +497,8 @@ irqreturn_t cvm_mmc_interrupt(int irq, void *dev_id)
>             (rsp_sts & MIO_EMM_RSP_STS_DMA_PEND))
>                 cleanup_dma(host, rsp_sts);
>

> +       mmc_request_done(slot->mmc, req);
>         host->current_req = NULL;
> -       req->done(req);

Flipping the order doesn't really matter here, as cvm_mmc_request() is
protected with the ->acquire_bus() lock.

However, I think it's good practise from the mmc core point of view,
to clear host->current_req prior to calling mmc_request_done(). Can
you please change this.

>
>  no_req_done:
>         if (host->dmar_fixup_done)
> @@ -699,8 +703,7 @@ static void cvm_mmc_dma_request(struct mmc_host *mmc,
>
>  error:
>         mrq->cmd->error = -EINVAL;
> -       if (mrq->done)
> -               mrq->done(mrq);
> +       mmc_request_done(mmc, mrq);
>         host->release_bus(host);
>  }
>

Kind regards
Uffe