| Message ID | 1386680168-5227-6-git-send-email-b29396@freescale.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tue, Dec 10, 2013 at 08:56:07PM +0800, Dong Aisheng wrote: > For host controllers using SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK, since the card > clock is changed dynamically for different cards, it does not make sense > to use the maximum host clock to calculate max_discard_to which may lead > the max_discard_to to be much smaller than its capbility and affect the card > discard performance a lot. > e.g. the host clock is 200Mhz, but the card is working on 50Mhz. Then the > max_discard_to is only 1/4 of its real capbility. > > In this patch, it uses the actual_clock to calculate the max_discard_to > dynamically as long as a new clock speed is set. > > Tested with a high speed SDHC card shows: > Originally: > mmc1: new high speed SDHC card at address aaaa > mmc1: calculated max. discard sectors 49152 for timeout 1355 ms > Now: > mmc1: new high speed SDHC card at address aaaa > mmc1: calculated max. discard sectors 712704 for timeout 5422 ms > The max_discard_sectors will increase a lot which will also improve discard > performance a lot. > > The one known limitation of this approach is that it does not cover the special > case for user changes the clock via sysfs, since the max_discard_to is only > initialised for one time during the mmc queue init. > > Signed-off-by: Dong Aisheng <b29396@freescale.com> > --- > drivers/mmc/host/sdhci.c | 27 +++++++++++++++++++++------ > 1 files changed, 21 insertions(+), 6 deletions(-) > > diff --git a/drivers/mmc/host/sdhci.c b/drivers/mmc/host/sdhci.c > index 4cc3bd6..9be8a79 100644 > --- a/drivers/mmc/host/sdhci.c > +++ b/drivers/mmc/host/sdhci.c > @@ -1143,14 +1143,14 @@ static void sdhci_set_clock(struct sdhci_host *host, unsigned int clock) > unsigned long timeout; > > if (clock && clock == host->clock) > - return; > + goto out; I do not quite understand this change. Why do we need to recalculate max_discard_to if the clock does not change since the last time that the function is called? Shawn > > host->mmc->actual_clock = 0; > > if (host->ops->set_clock) { > host->ops->set_clock(host, clock); > if (host->quirks & SDHCI_QUIRK_NONSTANDARD_CLOCK) > - return; > + goto out; > } > > sdhci_writew(host, 0, SDHCI_CLOCK_CONTROL); > @@ -1249,6 +1249,19 @@ clock_set: > > out: > host->clock = clock; > + > + /* update timeout_clk and max_discard_to once the SDCLK is changed */ > + if (host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK && clock) { > + host->timeout_clk = host->mmc->actual_clock ? > + host->mmc->actual_clock / 1000 : > + host->clock / 1000; > + if (host->ops->get_max_timeout) > + host->mmc->max_discard_to = > + host->ops->get_max_timeout(host); > + else > + host->mmc->max_discard_to = (1 << 27) / > + host->timeout_clk; > + } > } > > static inline void sdhci_update_clock(struct sdhci_host *host) > @@ -2939,10 +2952,12 @@ int sdhci_add_host(struct sdhci_host *host) > if (host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK) > host->timeout_clk = mmc->f_max / 1000; > > - if (host->ops->get_max_timeout) > - mmc->max_discard_to = host->ops->get_max_timeout(host); > - else > - mmc->max_discard_to = (1 << 27) / host->timeout_clk; > + if (!(host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK)) { > + if (host->ops->get_max_timeout) > + mmc->max_discard_to = host->ops->get_max_timeout(host); > + else > + mmc->max_discard_to = (1 << 27) / host->timeout_clk; > + } > > mmc->caps |= MMC_CAP_SDIO_IRQ | MMC_CAP_ERASE | MMC_CAP_CMD23; > > -- > 1.7.2.rc3 > > -- 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
On Wed, Dec 11, 2013 at 11:01 AM, Shawn Guo <shawn.guo@linaro.org> wrote: > On Tue, Dec 10, 2013 at 08:56:07PM +0800, Dong Aisheng wrote: >> For host controllers using SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK, since the card >> clock is changed dynamically for different cards, it does not make sense >> to use the maximum host clock to calculate max_discard_to which may lead >> the max_discard_to to be much smaller than its capbility and affect the card >> discard performance a lot. >> e.g. the host clock is 200Mhz, but the card is working on 50Mhz. Then the >> max_discard_to is only 1/4 of its real capbility. >> >> In this patch, it uses the actual_clock to calculate the max_discard_to >> dynamically as long as a new clock speed is set. >> >> Tested with a high speed SDHC card shows: >> Originally: >> mmc1: new high speed SDHC card at address aaaa >> mmc1: calculated max. discard sectors 49152 for timeout 1355 ms >> Now: >> mmc1: new high speed SDHC card at address aaaa >> mmc1: calculated max. discard sectors 712704 for timeout 5422 ms >> The max_discard_sectors will increase a lot which will also improve discard >> performance a lot. >> >> The one known limitation of this approach is that it does not cover the special >> case for user changes the clock via sysfs, since the max_discard_to is only >> initialised for one time during the mmc queue init. >> >> Signed-off-by: Dong Aisheng <b29396@freescale.com> >> --- >> drivers/mmc/host/sdhci.c | 27 +++++++++++++++++++++------ >> 1 files changed, 21 insertions(+), 6 deletions(-) >> >> diff --git a/drivers/mmc/host/sdhci.c b/drivers/mmc/host/sdhci.c >> index 4cc3bd6..9be8a79 100644 >> --- a/drivers/mmc/host/sdhci.c >> +++ b/drivers/mmc/host/sdhci.c >> @@ -1143,14 +1143,14 @@ static void sdhci_set_clock(struct sdhci_host *host, unsigned int clock) >> unsigned long timeout; >> >> if (clock && clock == host->clock) >> - return; >> + goto out; > > I do not quite understand this change. Why do we need to recalculate > max_discard_to if the clock does not change since the last time that > the function is called? > Good catch. It's safe to return directly here. Will update in v2. Regards Dong Aisheng > Shawn > >> >> host->mmc->actual_clock = 0; >> >> if (host->ops->set_clock) { >> host->ops->set_clock(host, clock); >> if (host->quirks & SDHCI_QUIRK_NONSTANDARD_CLOCK) >> - return; >> + goto out; >> } >> >> sdhci_writew(host, 0, SDHCI_CLOCK_CONTROL); >> @@ -1249,6 +1249,19 @@ clock_set: >> >> out: >> host->clock = clock; >> + >> + /* update timeout_clk and max_discard_to once the SDCLK is changed */ >> + if (host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK && clock) { >> + host->timeout_clk = host->mmc->actual_clock ? >> + host->mmc->actual_clock / 1000 : >> + host->clock / 1000; >> + if (host->ops->get_max_timeout) >> + host->mmc->max_discard_to = >> + host->ops->get_max_timeout(host); >> + else >> + host->mmc->max_discard_to = (1 << 27) / >> + host->timeout_clk; >> + } >> } >> >> static inline void sdhci_update_clock(struct sdhci_host *host) >> @@ -2939,10 +2952,12 @@ int sdhci_add_host(struct sdhci_host *host) >> if (host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK) >> host->timeout_clk = mmc->f_max / 1000; >> >> - if (host->ops->get_max_timeout) >> - mmc->max_discard_to = host->ops->get_max_timeout(host); >> - else >> - mmc->max_discard_to = (1 << 27) / host->timeout_clk; >> + if (!(host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK)) { >> + if (host->ops->get_max_timeout) >> + mmc->max_discard_to = host->ops->get_max_timeout(host); >> + else >> + mmc->max_discard_to = (1 << 27) / host->timeout_clk; >> + } >> >> mmc->caps |= MMC_CAP_SDIO_IRQ | MMC_CAP_ERASE | MMC_CAP_CMD23; >> >> -- >> 1.7.2.rc3 >> >> > > > _______________________________________________ > linux-arm-kernel mailing list > linux-arm-kernel@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-arm-kernel -- 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
On Tue, Dec 10, 2013 at 08:56:07PM +0800, Dong Aisheng wrote: > static inline void sdhci_update_clock(struct sdhci_host *host) > @@ -2939,10 +2952,12 @@ int sdhci_add_host(struct sdhci_host *host) > if (host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK) > host->timeout_clk = mmc->f_max / 1000; Since max_discard_to calculation below will not happen for SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK case, does it still make sense to keep the above code? Or put it in the other way, if we keep the above code and do not make the change below, will there be any problem besides the max_discard_to initialization plays for nothing? All in all, I'm just confused why we keep the above code and make the change below at the same time. Shawn > > - if (host->ops->get_max_timeout) > - mmc->max_discard_to = host->ops->get_max_timeout(host); > - else > - mmc->max_discard_to = (1 << 27) / host->timeout_clk; > + if (!(host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK)) { > + if (host->ops->get_max_timeout) > + mmc->max_discard_to = host->ops->get_max_timeout(host); > + else > + mmc->max_discard_to = (1 << 27) / host->timeout_clk; > + } > > mmc->caps |= MMC_CAP_SDIO_IRQ | MMC_CAP_ERASE | MMC_CAP_CMD23; > > -- > 1.7.2.rc3 > > -- 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
On Wed, Dec 11, 2013 at 12:05 PM, Shawn Guo <shawn.guo@linaro.org> wrote: > On Tue, Dec 10, 2013 at 08:56:07PM +0800, Dong Aisheng wrote: >> static inline void sdhci_update_clock(struct sdhci_host *host) >> @@ -2939,10 +2952,12 @@ int sdhci_add_host(struct sdhci_host *host) >> if (host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK) >> host->timeout_clk = mmc->f_max / 1000; > > Since max_discard_to calculation below will not happen for > SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK case, does it still make sense to > keep the above code? > Right, i missed to remove it. Will update in v2. > Or put it in the other way, if we keep the above code and do not make > the change below, will there be any problem besides the max_discard_to > initialization plays for nothing? > > All in all, I'm just confused why we keep the above code and make the > change below at the same time. > THe max_discard_to should be dynamically updated for SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK when changing the clock, so we can remove above lines. Regards Dong Aisheng > Shawn > >> >> - if (host->ops->get_max_timeout) >> - mmc->max_discard_to = host->ops->get_max_timeout(host); >> - else >> - mmc->max_discard_to = (1 << 27) / host->timeout_clk; >> + if (!(host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK)) { >> + if (host->ops->get_max_timeout) >> + mmc->max_discard_to = host->ops->get_max_timeout(host); >> + else >> + mmc->max_discard_to = (1 << 27) / host->timeout_clk; >> + } >> >> mmc->caps |= MMC_CAP_SDIO_IRQ | MMC_CAP_ERASE | MMC_CAP_CMD23; >> >> -- >> 1.7.2.rc3 >> >> > > > _______________________________________________ > linux-arm-kernel mailing list > linux-arm-kernel@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-arm-kernel -- 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/sdhci.c b/drivers/mmc/host/sdhci.c index 4cc3bd6..9be8a79 100644 --- a/drivers/mmc/host/sdhci.c +++ b/drivers/mmc/host/sdhci.c @@ -1143,14 +1143,14 @@ static void sdhci_set_clock(struct sdhci_host *host, unsigned int clock) unsigned long timeout; if (clock && clock == host->clock) - return; + goto out; host->mmc->actual_clock = 0; if (host->ops->set_clock) { host->ops->set_clock(host, clock); if (host->quirks & SDHCI_QUIRK_NONSTANDARD_CLOCK) - return; + goto out; } sdhci_writew(host, 0, SDHCI_CLOCK_CONTROL); @@ -1249,6 +1249,19 @@ clock_set: out: host->clock = clock; + + /* update timeout_clk and max_discard_to once the SDCLK is changed */ + if (host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK && clock) { + host->timeout_clk = host->mmc->actual_clock ? + host->mmc->actual_clock / 1000 : + host->clock / 1000; + if (host->ops->get_max_timeout) + host->mmc->max_discard_to = + host->ops->get_max_timeout(host); + else + host->mmc->max_discard_to = (1 << 27) / + host->timeout_clk; + } } static inline void sdhci_update_clock(struct sdhci_host *host) @@ -2939,10 +2952,12 @@ int sdhci_add_host(struct sdhci_host *host) if (host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK) host->timeout_clk = mmc->f_max / 1000; - if (host->ops->get_max_timeout) - mmc->max_discard_to = host->ops->get_max_timeout(host); - else - mmc->max_discard_to = (1 << 27) / host->timeout_clk; + if (!(host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK)) { + if (host->ops->get_max_timeout) + mmc->max_discard_to = host->ops->get_max_timeout(host); + else + mmc->max_discard_to = (1 << 27) / host->timeout_clk; + } mmc->caps |= MMC_CAP_SDIO_IRQ | MMC_CAP_ERASE | MMC_CAP_CMD23;
For host controllers using SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK, since the card clock is changed dynamically for different cards, it does not make sense to use the maximum host clock to calculate max_discard_to which may lead the max_discard_to to be much smaller than its capbility and affect the card discard performance a lot. e.g. the host clock is 200Mhz, but the card is working on 50Mhz. Then the max_discard_to is only 1/4 of its real capbility. In this patch, it uses the actual_clock to calculate the max_discard_to dynamically as long as a new clock speed is set. Tested with a high speed SDHC card shows: Originally: mmc1: new high speed SDHC card at address aaaa mmc1: calculated max. discard sectors 49152 for timeout 1355 ms Now: mmc1: new high speed SDHC card at address aaaa mmc1: calculated max. discard sectors 712704 for timeout 5422 ms The max_discard_sectors will increase a lot which will also improve discard performance a lot. The one known limitation of this approach is that it does not cover the special case for user changes the clock via sysfs, since the max_discard_to is only initialised for one time during the mmc queue init. Signed-off-by: Dong Aisheng <b29396@freescale.com> --- drivers/mmc/host/sdhci.c | 27 +++++++++++++++++++++------ 1 files changed, 21 insertions(+), 6 deletions(-)