| Message ID | 1532654046-11901-1-git-send-email-yamada.masahiro@socionext.com (mailing list archive) |
|---|---|
| State | Under Review |
| Delegated to: | Geert Uytterhoeven |
| Headers | show |
| Series | mmc: tmio: allow DMA request hook to return error status | expand |
Hi Yamada-san, On Fri, Jul 27, 2018 at 3:15 AM Masahiro Yamada <yamada.masahiro@socionext.com> wrote: > dma_request_chan() may return ERR_PTR(-EPROBE_DEFER), but > tmio_mmc_request_dma() cannot propagate it since it is a > void function. > > Change the return type to int so that the driver can retry > probing later in case the DMA-engine driver is probed after > the TMIO MMC driver. > > I moved the call for tmio_mmc_request_dma() up because it may > fail now. I also removed unneeded clearing of host->chan_{tx,rx} > because (struct tmio_mmc_host) is allocated by kzalloc(). > > Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com> Thanks for your patch! Isn't the idea to fall back to PIO if the DMA-engine driver is not found? Unfortunately there's no way to distinguish between "DMA-engine hasn't been probed yet" and "DMA-engine is not available" (e.g. CONFIG_RCAR_DMAC=n). In both cases, dma_request_chan() will return -EPROBE_DEFER. So if you treat this as an actual error, and propagate it, the following will happen: - In case 1, the MMC driver will be reprobed successfully later. - In case 2,the MMC driver will never succeed. Or am I missing something? Gr{oetje,eeting}s, Geert
Hi Geert, 2018-07-27 16:51 GMT+09:00 Geert Uytterhoeven <geert@linux-m68k.org>: > Hi Yamada-san, > > On Fri, Jul 27, 2018 at 3:15 AM Masahiro Yamada > <yamada.masahiro@socionext.com> wrote: >> dma_request_chan() may return ERR_PTR(-EPROBE_DEFER), but >> tmio_mmc_request_dma() cannot propagate it since it is a >> void function. >> >> Change the return type to int so that the driver can retry >> probing later in case the DMA-engine driver is probed after >> the TMIO MMC driver. >> >> I moved the call for tmio_mmc_request_dma() up because it may >> fail now. I also removed unneeded clearing of host->chan_{tx,rx} >> because (struct tmio_mmc_host) is allocated by kzalloc(). >> >> Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com> > > Thanks for your patch! > > Isn't the idea to fall back to PIO if the DMA-engine driver is not found? > > Unfortunately there's no way to distinguish between "DMA-engine hasn't been > probed yet" and "DMA-engine is not available" (e.g. CONFIG_RCAR_DMAC=n). > In both cases, dma_request_chan() will return -EPROBE_DEFER. > So if you treat this as an actual error, and propagate it, the following > will happen: > - In case 1, the MMC driver will be reprobed successfully later. > - In case 2,the MMC driver will never succeed. > > Or am I missing something? You are right. drivers/Makefile lists 'dma/' before 'mmc/', so this is not a real problem. I will let my driver fallback to PIO for any error. Thanks. > Gr{oetje,eeting}s, > > Geert > > -- > Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org > > In personal conversations with technical people, I call myself a hacker. But > when I'm talking to journalists I just say "programmer" or something like that. > -- Linus Torvalds > -- > To unsubscribe from this list: send the line "unsubscribe linux-mmc" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/mmc/host/renesas_sdhi_internal_dmac.c b/drivers/mmc/host/renesas_sdhi_internal_dmac.c index d032bd6..f3475a5 100644 --- a/drivers/mmc/host/renesas_sdhi_internal_dmac.c +++ b/drivers/mmc/host/renesas_sdhi_internal_dmac.c @@ -230,7 +230,7 @@ static void renesas_sdhi_internal_dmac_complete_tasklet_fn(unsigned long arg) spin_unlock_irq(&host->lock); } -static void +static int renesas_sdhi_internal_dmac_request_dma(struct tmio_mmc_host *host, struct tmio_mmc_data *pdata) { @@ -245,6 +245,8 @@ renesas_sdhi_internal_dmac_request_dma(struct tmio_mmc_host *host, tasklet_init(&host->dma_issue, renesas_sdhi_internal_dmac_issue_tasklet_fn, (unsigned long)host); + + return 0; } static void diff --git a/drivers/mmc/host/renesas_sdhi_sys_dmac.c b/drivers/mmc/host/renesas_sdhi_sys_dmac.c index 4bb46c4..3f806f5 100644 --- a/drivers/mmc/host/renesas_sdhi_sys_dmac.c +++ b/drivers/mmc/host/renesas_sdhi_sys_dmac.c @@ -359,15 +359,15 @@ static void renesas_sdhi_sys_dmac_issue_tasklet_fn(unsigned long priv) dma_async_issue_pending(chan); } -static void renesas_sdhi_sys_dmac_request_dma(struct tmio_mmc_host *host, - struct tmio_mmc_data *pdata) +static int renesas_sdhi_sys_dmac_request_dma(struct tmio_mmc_host *host, + struct tmio_mmc_data *pdata) { struct renesas_sdhi *priv = host_to_priv(host); /* We can only either use DMA for both Tx and Rx or not use it at all */ if (!host->pdev->dev.of_node && (!pdata->chan_priv_tx || !pdata->chan_priv_rx)) - return; + return 0; if (!host->chan_tx && !host->chan_rx) { struct resource *res = platform_get_resource(host->pdev, @@ -377,7 +377,7 @@ static void renesas_sdhi_sys_dmac_request_dma(struct tmio_mmc_host *host, int ret; if (!res) - return; + return 0; dma_cap_zero(mask); dma_cap_set(DMA_SLAVE, mask); @@ -389,7 +389,7 @@ static void renesas_sdhi_sys_dmac_request_dma(struct tmio_mmc_host *host, host->chan_tx); if (!host->chan_tx) - return; + return 0; cfg.direction = DMA_MEM_TO_DEV; cfg.dst_addr = res->start + @@ -433,7 +433,7 @@ static void renesas_sdhi_sys_dmac_request_dma(struct tmio_mmc_host *host, renesas_sdhi_sys_dmac_enable_dma(host, true); - return; + return 0; ebouncebuf: ecfgrx: @@ -443,6 +443,8 @@ static void renesas_sdhi_sys_dmac_request_dma(struct tmio_mmc_host *host, ecfgtx: dma_release_channel(host->chan_tx); host->chan_tx = NULL; + + return 0; } static void renesas_sdhi_sys_dmac_release_dma(struct tmio_mmc_host *host) diff --git a/drivers/mmc/host/tmio_mmc.h b/drivers/mmc/host/tmio_mmc.h index e7d6513..d8383d4 100644 --- a/drivers/mmc/host/tmio_mmc.h +++ b/drivers/mmc/host/tmio_mmc.h @@ -115,8 +115,7 @@ struct tmio_mmc_host; struct tmio_mmc_dma_ops { void (*start)(struct tmio_mmc_host *host, struct mmc_data *data); void (*enable)(struct tmio_mmc_host *host, bool enable); - void (*request)(struct tmio_mmc_host *host, - struct tmio_mmc_data *pdata); + int (*request)(struct tmio_mmc_host *host, struct tmio_mmc_data *pdata); void (*release)(struct tmio_mmc_host *host); void (*abort)(struct tmio_mmc_host *host); void (*dataend)(struct tmio_mmc_host *host); diff --git a/drivers/mmc/host/tmio_mmc_core.c b/drivers/mmc/host/tmio_mmc_core.c index 3cb554c..09c2d0c 100644 --- a/drivers/mmc/host/tmio_mmc_core.c +++ b/drivers/mmc/host/tmio_mmc_core.c @@ -65,15 +65,13 @@ static inline void tmio_mmc_enable_dma(struct tmio_mmc_host *host, bool enable) host->dma_ops->enable(host, enable); } -static inline void tmio_mmc_request_dma(struct tmio_mmc_host *host, - struct tmio_mmc_data *pdata) +static inline int tmio_mmc_request_dma(struct tmio_mmc_host *host, + struct tmio_mmc_data *pdata) { - if (host->dma_ops) { - host->dma_ops->request(host, pdata); - } else { - host->chan_tx = NULL; - host->chan_rx = NULL; - } + if (host->dma_ops) + return host->dma_ops->request(host, pdata); + + return 0; } static inline void tmio_mmc_release_dma(struct tmio_mmc_host *host) @@ -1218,6 +1216,10 @@ int tmio_mmc_host_probe(struct tmio_mmc_host *_host) return ret; } + ret = tmio_mmc_request_dma(_host, pdata); + if (ret) + return ret; + mmc->caps |= MMC_CAP_4_BIT_DATA | pdata->capabilities; mmc->caps2 |= pdata->capabilities2; mmc->max_segs = pdata->max_segs ? : 32; @@ -1286,9 +1288,6 @@ int tmio_mmc_host_probe(struct tmio_mmc_host *_host) INIT_DELAYED_WORK(&_host->delayed_reset_work, tmio_mmc_reset_work); INIT_WORK(&_host->done, tmio_mmc_done_work); - /* See if we also get DMA */ - tmio_mmc_request_dma(_host, pdata); - pm_runtime_set_active(&pdev->dev); pm_runtime_set_autosuspend_delay(&pdev->dev, 50); pm_runtime_use_autosuspend(&pdev->dev);
dma_request_chan() may return ERR_PTR(-EPROBE_DEFER), but tmio_mmc_request_dma() cannot propagate it since it is a void function. Change the return type to int so that the driver can retry probing later in case the DMA-engine driver is probed after the TMIO MMC driver. I moved the call for tmio_mmc_request_dma() up because it may fail now. I also removed unneeded clearing of host->chan_{tx,rx} because (struct tmio_mmc_host) is allocated by kzalloc(). Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com> --- drivers/mmc/host/renesas_sdhi_internal_dmac.c | 4 +++- drivers/mmc/host/renesas_sdhi_sys_dmac.c | 14 ++++++++------ drivers/mmc/host/tmio_mmc.h | 3 +-- drivers/mmc/host/tmio_mmc_core.c | 21 ++++++++++----------- 4 files changed, 22 insertions(+), 20 deletions(-)