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 <[email protected]> --- drivers/mmc/host/sdhci.c | 47 +++++++++++++++++++++++++++------------------ 1 files changed, 28 insertions(+), 19 deletions(-) diff --git a/drivers/mmc/host/sdhci.c b/drivers/mmc/host/sdhci.c index 47ecf50..a104bb1 100644 --- a/drivers/mmc/host/sdhci.c +++ b/drivers/mmc/host/sdhci.c @@ -1155,7 +1155,7 @@ static void sdhci_set_clock(struct sdhci_host *host, unsigned int clock) 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); @@ -1254,6 +1254,16 @@ 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; + host->mmc->max_discard_to = host->ops->get_max_timeout_count ? + host->ops->get_max_timeout_count(host) : 1 << 27; + host->mmc->max_discard_to /= host->timeout_clk; + } } static inline void sdhci_update_clock(struct sdhci_host *host) @@ -2933,27 +2943,26 @@ int sdhci_add_host(struct sdhci_host *host) } else mmc->f_min = host->max_clk / SDHCI_MAX_DIV_SPEC_200; - host->timeout_clk = - (caps[0] & SDHCI_TIMEOUT_CLK_MASK) >> SDHCI_TIMEOUT_CLK_SHIFT; - if (host->timeout_clk == 0) { - if (host->ops->get_timeout_clock) { - host->timeout_clk = host->ops->get_timeout_clock(host); - } else if (!(host->quirks & - SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK)) { - pr_err("%s: Hardware doesn't specify timeout clock " - "frequency.\n", mmc_hostname(mmc)); - return -ENODEV; + if (!(host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK)) { + host->timeout_clk = (caps[0] & SDHCI_TIMEOUT_CLK_MASK) >> + SDHCI_TIMEOUT_CLK_SHIFT; + if (host->timeout_clk == 0) { + if (host->ops->get_timeout_clock) { + host->timeout_clk = + host->ops->get_timeout_clock(host); + } else { + pr_err("%s: Hardware doesn't specify timeout" + "clock frequency.\n", mmc_hostname(mmc)); + return -ENODEV; + } } - } - if (caps[0] & SDHCI_TIMEOUT_CLK_UNIT) - host->timeout_clk *= 1000; - - if (host->quirks & SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK) - host->timeout_clk = mmc->f_max / 1000; + if (caps[0] & SDHCI_TIMEOUT_CLK_UNIT) + host->timeout_clk *= 1000; - mmc->max_discard_to = host->ops->get_max_timeout_count ? + mmc->max_discard_to = host->ops->get_max_timeout_count ? host->ops->get_max_timeout_count(host) : 1 << 27; - mmc->max_discard_to /= host->timeout_clk; + mmc->max_discard_to /= 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 [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
