Re: [PATCH 1/2] omap: hsmmc: Normalize dma cleanup operations
On 14 September 2011 15:40, S, Venkatraman wrote: > On Tue, Sep 13, 2011 at 1:26 AM, Per Forlin wrote: >> On 1 September 2011 21:05, Venkatraman S wrote: >>> Reuse omap_hsmmc_dma_cleanup even for normal dma teardown in >>> omap_hsmmc_dma_cb. Consolidate multiple points of dma unmap into a >>> single location in post_req function, to prevent double unmapping. >> It's optional to use pre_req() and post_req(). The SDIO framework >> doesn't utilise these hooks. For instance this wont work together with >> SDIO-wlan on the pandaboard. >> If pre_req() has been issued it's fine to defer dma_unmap() until >> post_req(). If pre_req() is not called the driver should call >> dma_unmap() directly. >> > > In that case, can the actual 'request' function just call pre_req and post_req > (at the beginning and at the end), if host_cookie is not set ? > Which request() function do your refer to? The mmc_host_ops.request() returns before the transfer is finished. Apart from that, I would say yes. It's nice to have the same code path for valid cookie and invalid cookie scenario. Note that the host_cookie could be set but with an invalid value. host driver: 1. validate cookie 2. if invalid cookie call pre_req() 3. run request 4. done -> if (has called pre_req() #2) -> call post_req(). A general fix would be to put this logic in mmc_wait_for_req() in core.c, but at this level it's not possible to validate the cookie. In the current implementation the host driver controls the cookie value. /Per -- To unsubscribe from this list: send the line "unsubscribe linux-mmc" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 1/2] omap: hsmmc: Normalize dma cleanup operations
On Tue, Sep 13, 2011 at 1:26 AM, Per Forlin wrote: > On 1 September 2011 21:05, Venkatraman S wrote: >> Reuse omap_hsmmc_dma_cleanup even for normal dma teardown in >> omap_hsmmc_dma_cb. Consolidate multiple points of dma unmap into a >> single location in post_req function, to prevent double unmapping. > It's optional to use pre_req() and post_req(). The SDIO framework > doesn't utilise these hooks. For instance this wont work together with > SDIO-wlan on the pandaboard. > If pre_req() has been issued it's fine to defer dma_unmap() until > post_req(). If pre_req() is not called the driver should call > dma_unmap() directly. > In that case, can the actual 'request' function just call pre_req and post_req (at the beginning and at the end), if host_cookie is not set ? -- To unsubscribe from this list: send the line "unsubscribe linux-mmc" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 1/2] omap: hsmmc: Normalize dma cleanup operations
On Tue, Sep 13, 2011 at 1:26 AM, Per Forlin wrote: > On 1 September 2011 21:05, Venkatraman S wrote: >> Reuse omap_hsmmc_dma_cleanup even for normal dma teardown in >> omap_hsmmc_dma_cb. Consolidate multiple points of dma unmap into a >> single location in post_req function, to prevent double unmapping. > It's optional to use pre_req() and post_req(). The SDIO framework > doesn't utilise these hooks. For instance this wont work together with > SDIO-wlan on the pandaboard. > If pre_req() has been issued it's fine to defer dma_unmap() until > post_req(). If pre_req() is not called the driver should call > dma_unmap() directly. > Thanks for the clarification. I'll redo the fix. -- To unsubscribe from this list: send the line "unsubscribe linux-mmc" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 1/2] omap: hsmmc: Normalize dma cleanup operations
On 1 September 2011 21:05, Venkatraman S wrote: > Reuse omap_hsmmc_dma_cleanup even for normal dma teardown in > omap_hsmmc_dma_cb. Consolidate multiple points of dma unmap into a > single location in post_req function, to prevent double unmapping. It's optional to use pre_req() and post_req(). The SDIO framework doesn't utilise these hooks. For instance this wont work together with SDIO-wlan on the pandaboard. If pre_req() has been issued it's fine to defer dma_unmap() until post_req(). If pre_req() is not called the driver should call dma_unmap() directly. Regards, Per -- To unsubscribe from this list: send the line "unsubscribe linux-mmc" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 1/2] omap: hsmmc: Normalize dma cleanup operations
On Fri, Sep 2, 2011 at 12:35 AM, Venkatraman S wrote: > Reuse omap_hsmmc_dma_cleanup even for normal dma teardown in > omap_hsmmc_dma_cb. Consolidate multiple points of dma unmap into a > single location in post_req function, to prevent double unmapping. > > Signed-off-by: Venkatraman S > --- > drivers/mmc/host/omap_hsmmc.c | 20 +--- > 1 files changed, 5 insertions(+), 15 deletions(-) > > diff --git a/drivers/mmc/host/omap_hsmmc.c b/drivers/mmc/host/omap_hsmmc.c > index 21e4a79..5b7776c 100644 > --- a/drivers/mmc/host/omap_hsmmc.c > +++ b/drivers/mmc/host/omap_hsmmc.c > @@ -999,7 +999,8 @@ static void omap_hsmmc_dma_cleanup(struct omap_hsmmc_host > *host, int errno) > { > int dma_ch; > > - host->data->error = errno; > + if (host->data) The condition check above becomes redundant since dma_cleanup is called during errors in omap_hsmmc_do_irq. Better have this check in the dma_cb where dma_cleanup is newly added. > + host->data->error = errno; > > spin_lock(&host->irq_lock); > dma_ch = host->dma_ch; > @@ -1007,12 +1008,8 @@ static void omap_hsmmc_dma_cleanup(struct > omap_hsmmc_host *host, int errno) > spin_unlock(&host->irq_lock); > > if (host->use_dma && dma_ch != -1) { > - dma_unmap_sg(mmc_dev(host->mmc), host->data->sg, > - host->data->sg_len, > - omap_hsmmc_get_dma_dir(host, host->data)); > omap_free_dma(dma_ch); > } > - host->data = NULL; No need to remove the above one. > } > > /* > @@ -1370,7 +1367,7 @@ static void omap_hsmmc_dma_cb(int lch, u16 ch_status, > void *cb_data) > { > struct omap_hsmmc_host *host = cb_data; > struct mmc_data *data; > - int dma_ch, req_in_progress; > + int req_in_progress; > > if (!(ch_status & OMAP_DMA_BLOCK_IRQ)) { > dev_warn(mmc_dev(host->mmc), "unexpected dma status %x\n", > @@ -1394,16 +1391,9 @@ static void omap_hsmmc_dma_cb(int lch, u16 ch_status, > void *cb_data) > return; > } > > - if (!data->host_cookie) > - dma_unmap_sg(mmc_dev(host->mmc), data->sg, data->sg_len, > - omap_hsmmc_get_dma_dir(host, data)); > - > req_in_progress = host->req_in_progress; > - dma_ch = host->dma_ch; > - host->dma_ch = -1; > spin_unlock(&host->irq_lock); > - > - omap_free_dma(dma_ch); > + omap_hsmmc_dma_cleanup(host, 0); > > /* If DMA has finished after TC, complete the request */ > if (!req_in_progress) { > @@ -1575,7 +1565,7 @@ static void omap_hsmmc_post_req(struct mmc_host *mmc, > struct mmc_request *mrq, > struct omap_hsmmc_host *host = mmc_priv(mmc); > struct mmc_data *data = mrq->data; > > - if (host->use_dma) { Here also , the above conditional check is required. > + if (data->host_cookie) { > dma_unmap_sg(mmc_dev(host->mmc), data->sg, data->sg_len, > omap_hsmmc_get_dma_dir(host, data)); > data->host_cookie = 0; > -- > 1.7.1 > > -- > To unsubscribe from this list: send the line "unsubscribe linux-omap" in > the body of a message to majord...@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > -- To unsubscribe from this list: send the line "unsubscribe linux-mmc" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html