linux-kernel - Re: [RFC PATCH V3 13/21] mmc: sdhci: UHS-II support, skip signal_voltage

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20200916005213.GA2942982@laputa>
Date:   Wed, 16 Sep 2020 09:52:13 +0900
From:   AKASHI Takahiro <takahiro.akashi@...aro.org>
To:     Ben Chuang <benchuanggli@...il.com>
Cc:     Adrian Hunter <adrian.hunter@...el.com>,
        Ulf Hansson <ulf.hansson@...aro.org>,
        linux-mmc <linux-mmc@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Ben Chuang <ben.chuang@...esyslogic.com.tw>,
        greg.tu@...esyslogic.com.tw
Subject: Re: [RFC PATCH V3 13/21] mmc: sdhci: UHS-II support, skip
 signal_voltage_switch()

On Tue, Sep 15, 2020 at 07:36:14PM +0800, Ben Chuang wrote:
> Hi Takahiro,
> 
> On Tue, Sep 15, 2020 at 2:03 PM AKASHI Takahiro
> <takahiro.akashi@...aro.org> wrote:
> >
> > Ben, Adrian,
> >
> > On Mon, Sep 14, 2020 at 11:08:14AM +0300, Adrian Hunter wrote:
> > > On 14/09/20 9:40 am, AKASHI Takahiro wrote:
> > > > Adrian,
> > > >
> > > > On Fri, Aug 21, 2020 at 05:09:01PM +0300, Adrian Hunter wrote:
> > > >> On 10/07/20 2:11 pm, Ben Chuang wrote:
> > > >>> From: AKASHI Takahiro <takahiro.akashi@...aro.org>
> > > >>>
> > > >>> sdhci_start_signal_voltage_switch() should be called only in UHS-I mode,
> > > >>> and not for UHS-II mode.
> > > >>>
> > > >>> Signed-off-by: Ben Chuang <ben.chuang@...esyslogic.com.tw>
> > > >>> Signed-off-by: AKASHI Takahiro <takahiro.akashi@...aro.org>
> > > >>> ---
> > > >>>  drivers/mmc/host/sdhci.c | 7 ++++++-
> > > >>>  1 file changed, 6 insertions(+), 1 deletion(-)
> > > >>>
> > > >>> diff --git a/drivers/mmc/host/sdhci.c b/drivers/mmc/host/sdhci.c
> > > >>> index 5511649946b9..7f2537648a08 100644
> > > >>> --- a/drivers/mmc/host/sdhci.c
> > > >>> +++ b/drivers/mmc/host/sdhci.c
> > > >>> @@ -2623,8 +2623,13 @@ int sdhci_start_signal_voltage_switch(struct mmc_host *mmc,
> > > >>>   /*
> > > >>>    * Signal Voltage Switching is only applicable for Host Controllers
> > > >>>    * v3.00 and above.
> > > >>> +  * But for UHS2, the signal voltage is supplied by vdd2 which is
> > > >>> +  * already 1.8v so no voltage switch required.
> >
> > I have been confused with this comment.
> > (I know it came from the original Intel code, not from Ben.)
> >
> > If this comment is true,
> >
> > > >>>    */
> > > >>> - if (host->version < SDHCI_SPEC_300)
> > > >>> + if (host->version < SDHCI_SPEC_300 ||
> > > >>> +     (IS_ENABLED(CONFIG_MMC_SDHCI_UHS2) &&
> > > >>> +      host->version >= SDHCI_SPEC_400 &&
> > > >>> +      host->mmc->flags & MMC_UHS2_SUPPORT))
> >
> > the condition above must be wrong since 'flags & MMC_UHS2_SUPPORT'
> > is one of capabilities for a host controller, not a card
> > while the selection of voltage depends on a card type.
> 
> The flag MMC_UHS2_SUPPORT is set at the beginning of mmc_uhs2_rescan_try_freq().
> In UHS-II flow, it stays set.
> If the attempt to UHS-II fails finally, it will be unset.

Right, but MMC_UHS2_SUPPORT is also set, at least initially,
in sdhci_uhs2_add_host(). It is confusing, isn't it?

As we discussed before, any card-specific properties, like UHS-II mode,
should be placed in a card structure, not a host structure.

> >
> > So I wonder why this code still works.
> > I guess that it is because set_signal_voltage(), or other variant functions,
> > will never be called for UHS-II cards under the current implementation.
> >
> > Looking at mmc_sd_init_card(), we have added some hack:
> > mmc_sd_init_card()
> > {
> >         ...
> >         /* For UHS2, skip the UHS-I initialization. */
> >         if ((host->flags & MMC_UHS2_SUPPORT) &&
> >             (host->flags & MMC_UHS2_INITIALIZED))
> >                 goto done;
> >         ...
> >                 if (mmc_sd_card_using_v18(card)) {
> >                         if (mmc_host_set_uhs_voltage(host) ||
> >                             mmc_sd_init_uhs_card(card)) {
> >                 ...
> > }
> >
> > Ben, can you confirm this?
> > (There is another callsite of mmc_host_set_uhs_voltage() though.)
> 
> UHS-II cards use differential signals and don't need to signal voltage switch.
> But the main task is to set the parameters of UHS-II card interface.

Whoever sets MMC_UHS2_SUPPORT (and MMC_UHS2_INITIALIZED), my assertion above
(mmc_host_set_uhs_voltage, and hence [sdhci_]start_signal_voltage_switch(), is
never called for UHS-II cards) will be valid, isn't it?

-Takahiro Akashi

> >
> > > >> Please look at hooking ->start_signal_voltage_switch() instead
> > > >
> > > > Do you mean that you want every platform driver who wants to support UHS-II
> > > > to set NULL to start_signal_voltage_switch hook even if this hack is
> > > > platform agnostic?
> > >
> > > No, I see UHS-II as a separate layer i.e.
> > >
> > >  UHS-II host controller driver
> > >   |   |
> > >   |   v
> > >   |   sdhci-uhs2 e.g. sdhci_uhs2_start_signal_voltage_switch
> > >   |   |
> > >   v   v
> > >   sdhci e.g. sdhci_start_signal_voltage_switch
> > >
> > > Most things should go through sdhci-uhs2 but not nessarily everything.
> >
> > What I meant by my previous comment is that we don't have to
> > call any function, sdhci_uhs2_start_signal_voltage_switch in above example,
> > for UHS-II cards in any case since it is always simply empty.
> >
> > -Takahiro Akashi