From: Adrian Hunter <adrian.hunter@intel.com>
To: Lai Jason <jasonlai.genesyslogic@gmail.com>
Cc: "Ulf Hansson" <ulf.hansson@linaro.org>,
linux-mmc <linux-mmc@vger.kernel.org>,
"Ben Chuang" <ben.chuang@genesyslogic.com.tw>,
"GregTu[杜啟軒]" <greg.tu@genesyslogic.com.tw>,
SeanHY.chen@genesyslogic.com.tw,
"Jason Lai" <jason.lai@genesyslogic.com.tw>,
victor.shih@genesyslogic.com.tw,
"Renius Chen" <reniuschengl@gmail.com>
Subject: Re: [RESEND] mmc: sdhci-pci-gli: Improve Random 4K Read Performance of GL9763E
Date: Mon, 30 May 2022 07:46:39 +0300 [thread overview]
Message-ID: <521115fb-9bbc-0976-2cae-24b9c3e27e3a@intel.com> (raw)
In-Reply-To: <CAG0XXUHCd=fsR37KRbDjh4O+zn7bZfyQ0syDZvR7-Hk3RZH4EA@mail.gmail.com>
On 30/05/22 06:11, Lai Jason wrote:
> Hi Adrian,
>
> Please see my reply below.
>
> On Fri, May 27, 2022 at 4:54 PM Adrian Hunter <adrian.hunter@intel.com> wrote:
>>
>> On 26/05/22 10:08, jasonlai.genesyslogic@gmail.com wrote:
>>> From: Jason Lai <jasonlai.genesyslogic@gmail.com>
>>>
>>> Resend this patch due to code base updated to 5.18.0-rc3.
>>>
>>> This patch is based on patch [1] and remove data transfer length checking.
>>>
>>> Due to flaws in hardware design, GL9763E takes long time to exit from L1
>>> state. The I/O performance will suffer severe impact if it often enter and
>>> and exit L1 state.
>>>
>>> Unfortunately, entering and exiting L1 state is signal handshake in
>>> physical layer, software knows nothiong about it. The only way to stop
>>> entering L1 state is to disable hardware LPM negotiation on GL9763E.
>>>
>>> To improve read performance and take battery life into account, we reject
>>> L1 negotiation while executing MMC_READ_MULTIPLE_BLOCK command and enable L1
>>> negotiation again when receiving non-MMC_READ_MULTIPLE_BLOCK command.
>>>
>>> [1]
>>> https://patchwork.kernel.org/project/linux-mmc/list/?series=510801&archive
>>> =both
>>
>> Really needs Ulf's response, but a minor comment below.
>>
>>>
>>> Signed-off-by: Renius Chen <reniuschengl@gmail.com>
>>> Signed-off-by: Jason Lai <jason.lai@genesyslogic.com.tw>
>>> ---
>>> drivers/mmc/host/sdhci-pci-gli.c | 60 +++++++++++++++++++++++++++++++-
>>> 1 file changed, 59 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/mmc/host/sdhci-pci-gli.c b/drivers/mmc/host/sdhci-pci-gli.c
>>> index d09728c37d03..86200b73c0b0 100644
>>> --- a/drivers/mmc/host/sdhci-pci-gli.c
>>> +++ b/drivers/mmc/host/sdhci-pci-gli.c
>>> @@ -95,9 +95,12 @@
>>> #define PCIE_GLI_9763E_SCR 0x8E0
>>> #define GLI_9763E_SCR_AXI_REQ BIT(9)
>>>
>>> +#define PCIE_GLI_9763E_CFG 0x8A0
>>> +#define GLI_9763E_CFG_LPSN_DIS BIT(12)
>>> +
>>> #define PCIE_GLI_9763E_CFG2 0x8A4
>>> #define GLI_9763E_CFG2_L1DLY GENMASK(28, 19)
>>> -#define GLI_9763E_CFG2_L1DLY_MID 0x54
>>> +#define GLI_9763E_CFG2_L1DLY_MID 0x54 // Set L1 entry delay time to 21us
>>>
>>> #define PCIE_GLI_9763E_MMC_CTRL 0x960
>>> #define GLI_9763E_HS400_SLOW BIT(3)
>>> @@ -144,6 +147,10 @@
>>>
>>> #define GLI_MAX_TUNING_LOOP 40
>>>
>>> +struct gli_host {
>>> + bool lpm_negotiation_enabled;
>>> +};
>>> +
>>> /* Genesys Logic chipset */
>>> static inline void gl9750_wt_on(struct sdhci_host *host)
>>> {
>>> @@ -818,6 +825,53 @@ static void sdhci_gl9763e_dumpregs(struct mmc_host *mmc)
>>> sdhci_dumpregs(mmc_priv(mmc));
>>> }
>>>
>>> +static void gl9763e_set_low_power_negotiation(struct sdhci_pci_slot *slot, bool enable)
>>> +{
>>> + struct pci_dev *pdev = slot->chip->pdev;
>>> + u32 value;
>>> +
>>> + pci_read_config_dword(pdev, PCIE_GLI_9763E_VHS, &value);
>>> + value &= ~GLI_9763E_VHS_REV;
>>> + value |= FIELD_PREP(GLI_9763E_VHS_REV, GLI_9763E_VHS_REV_W);
>>> + pci_write_config_dword(pdev, PCIE_GLI_9763E_VHS, value);
>>> +
>>> + pci_read_config_dword(pdev, PCIE_GLI_9763E_CFG, &value);
>>> +
>>> + if (enable)
>>> + value &= ~GLI_9763E_CFG_LPSN_DIS;
>>> + else
>>> + value |= GLI_9763E_CFG_LPSN_DIS;
>>> +
>>> + pci_write_config_dword(pdev, PCIE_GLI_9763E_CFG, value);
>>> +
>>> + pci_read_config_dword(pdev, PCIE_GLI_9763E_VHS, &value);
>>> + value &= ~GLI_9763E_VHS_REV;
>>> + value |= FIELD_PREP(GLI_9763E_VHS_REV, GLI_9763E_VHS_REV_R);
>>> + pci_write_config_dword(pdev, PCIE_GLI_9763E_VHS, value);
>>> +}
>>> +
>>> +static void gl9763e_request(struct mmc_host *mmc, struct mmc_request *mrq)
>>> +{
>>> + struct sdhci_host *host = mmc_priv(mmc);
>>> + struct mmc_command *cmd;
>>> + struct sdhci_pci_slot *slot = sdhci_priv(host);
>>> + struct gli_host *gli_host = sdhci_pci_priv(slot);
>>> +
>>> + cmd = mrq->cmd;
>>> +
>>> + if (cmd && (cmd->opcode == MMC_READ_MULTIPLE_BLOCK) && gli_host->lpm_negotiation_enabled) {
>>> + gl9763e_set_low_power_negotiation(slot, false);
>>> + gli_host->lpm_negotiation_enabled = false;
>>> + } else {
>>> + if (gli_host->lpm_negotiation_enabled == false) {
>>
>> Is this logic right? Wouldn't it also get here with
>
> No, the logic is wrong. The original intention of my design is keeping LPM
> negotiation disabled from arriving of READ_MULTIPLE_BLOCK command to
> arriving of non-READ_MULTIPLE_BLOCK command. So The piece of code should be
> written as below:
> if (cmd && (cmd->opcode == MMC_READ_MULTIPLE_BLOCK)) {
> if (gli_host->lpm_negotiation_enabled) {
> gl9763e_set_low_power_negotiation(slot, false);
> gli_host->lpm_negotiation_enabled = false;
> }
> } else {
> if (gli_host->lpm_negotiation_enabled == false) {
> gl9763e_set_low_power_negotiation(slot, true);
> gli_host->lpm_negotiation_enabled = true;
> }
> }
>
> Am I correct?
Looks better. You might want to consider a wrapper function like:
static void gl9763e_set_lpm_negotiation(struct sdhci_pci_slot *slot, bool enable)
{
if (gli_host->lpm_negotiation_enabled == enable)
return;
gli_host->lpm_negotiation_enabled = enable;
gl9763e_set_low_power_negotiation(slot, enable);
}
>
> regards,
> Jason Lai
>
>> cmd->opcode == MMC_READ_MULTIPLE_BLOCK &&
>> gli_host->lpm_negotiation_enabled == false
>>
>> and then you don't want the following?
>>
>>> + gl9763e_set_low_power_negotiation(slot, true);
>>> + gli_host->lpm_negotiation_enabled = true;
>>> + }
>>> + }
>>> +
>>> + sdhci_request(mmc, mrq);
>>> +}
>>> +
>>> static void sdhci_gl9763e_cqe_pre_enable(struct mmc_host *mmc)
>>> {
>>> struct cqhci_host *cq_host = mmc->cqe_private;
>>> @@ -1016,6 +1070,9 @@ static int gli_probe_slot_gl9763e(struct sdhci_pci_slot *slot)
>>> gli_pcie_enable_msi(slot);
>>> host->mmc_host_ops.hs400_enhanced_strobe =
>>> gl9763e_hs400_enhanced_strobe;
>>> +
>>> + host->mmc_host_ops.request = gl9763e_request;
>>> +
>>> gli_set_gl9763e(slot);
>>> sdhci_enable_v4_mode(host);
>>>
>>> @@ -1109,4 +1166,5 @@ const struct sdhci_pci_fixes sdhci_gl9763e = {
>>> .allow_runtime_pm = true,
>>> #endif
>>> .add_host = gl9763e_add_host,
>>> + .priv_size = sizeof(struct gli_host),
>>> };
>>
prev parent reply other threads:[~2022-05-30 4:46 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-05-26 7:08 [RESEND] mmc: sdhci-pci-gli: Improve Random 4K Read Performance of GL9763E jasonlai.genesyslogic
2022-05-27 8:54 ` Adrian Hunter
2022-05-30 3:11 ` Lai Jason
2022-05-30 4:46 ` Adrian Hunter [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=521115fb-9bbc-0976-2cae-24b9c3e27e3a@intel.com \
--to=adrian.hunter@intel.com \
--cc=SeanHY.chen@genesyslogic.com.tw \
--cc=ben.chuang@genesyslogic.com.tw \
--cc=greg.tu@genesyslogic.com.tw \
--cc=jason.lai@genesyslogic.com.tw \
--cc=jasonlai.genesyslogic@gmail.com \
--cc=linux-mmc@vger.kernel.org \
--cc=reniuschengl@gmail.com \
--cc=ulf.hansson@linaro.org \
--cc=victor.shih@genesyslogic.com.tw \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox