| 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
| ||
|
Message-ID: <20191022171505.GK20212@google.com>
Date: Tue, 22 Oct 2019 10:15:05 -0700
From: Matthias Kaehlcke <mka@...omium.org>
To: Balakrishna Godavarthi <bgodavar@...eaurora.org>
Cc: Bjorn Andersson <bjorn.andersson@...aro.org>,
Marcel Holtmann <marcel@...tmann.org>,
Johan Hedberg <johan.hedberg@...il.com>,
linux-bluetooth@...r.kernel.org, linux-kernel@...r.kernel.org,
Jeffrey Hugo <jeffrey.l.hugo@...il.com>,
linux-arm-msm@...r.kernel.org,
linux-bluetooth-owner@...r.kernel.org, hemantg@...eaurora.org,
Harish Bandi <c-hbandi@...eaurora.org>,
Stephen Boyd <swboyd@...omium.org>
Subject: Re: [PATCH 2/4] Bluetooth: hci_qca: Don't vote for specific voltage
On Tue, Oct 22, 2019 at 11:35:43AM +0530, Balakrishna Godavarthi wrote:
> Hi Matthias, Bjorn andresson,
>
> On 2019-10-21 12:07, Harish Bandi wrote:
> > + Bala
> >
> > On 2019-10-18 23:52, Matthias Kaehlcke wrote:
> > > On Thu, Oct 17, 2019 at 10:24:02PM -0700, Bjorn Andersson wrote:
> > > > Devices with specific voltage requirements should not request voltage
> > > > from the driver, but instead rely on the system configuration to
> > > > define
> > > > appropriate voltages for each rail.
> > > >
> > > > This ensures that PMIC and board variations are accounted for,
> > > > something
> > > > that the 0.1V range in the hci_qca driver currently tries to address.
> > > > But on the Lenovo Yoga C630 (with wcn3990) vddch0 is 3.1V, which
> > > > means
> > > > the driver will fail to set the voltage.
> > > >
> > > > Signed-off-by: Bjorn Andersson <bjorn.andersson@...aro.org>
> > > > ---
> > > > drivers/bluetooth/hci_qca.c | 26 ++++++++------------------
> > > > 1 file changed, 8 insertions(+), 18 deletions(-)
> > > >
> > > > diff --git a/drivers/bluetooth/hci_qca.c
> > > > b/drivers/bluetooth/hci_qca.c
> > > > index c07c529b0d81..54aafcc69d06 100644
> > > > --- a/drivers/bluetooth/hci_qca.c
> > > > +++ b/drivers/bluetooth/hci_qca.c
> > > > @@ -130,8 +130,6 @@ enum qca_speed_type {
> > > > */
> > > > struct qca_vreg {
> > > > const char *name;
> > > > - unsigned int min_uV;
> > > > - unsigned int max_uV;
> > > > unsigned int load_uA;
> > > > };
> > > >
> > > > @@ -1332,10 +1330,10 @@ static const struct hci_uart_proto
> > > > qca_proto = {
> > > > static const struct qca_vreg_data qca_soc_data_wcn3990 = {
> > > > .soc_type = QCA_WCN3990,
> > > > .vregs = (struct qca_vreg []) {
> > > > - { "vddio", 1800000, 1900000, 15000 },
> > > > - { "vddxo", 1800000, 1900000, 80000 },
> > > > - { "vddrf", 1300000, 1350000, 300000 },
> > > > - { "vddch0", 3300000, 3400000, 450000 },
> > > > + { "vddio", 15000 },
> > > > + { "vddxo", 80000 },
> > > > + { "vddrf", 300000 },
> > > > + { "vddch0", 450000 },
> > > > },
> > > > .num_vregs = 4,
> > > > };
> > > > @@ -1343,10 +1341,10 @@ static const struct qca_vreg_data
> > > > qca_soc_data_wcn3990 = {
> > > > static const struct qca_vreg_data qca_soc_data_wcn3998 = {
> > > > .soc_type = QCA_WCN3998,
> > > > .vregs = (struct qca_vreg []) {
> > > > - { "vddio", 1800000, 1900000, 10000 },
> > > > - { "vddxo", 1800000, 1900000, 80000 },
> > > > - { "vddrf", 1300000, 1352000, 300000 },
> > > > - { "vddch0", 3300000, 3300000, 450000 },
> > > > + { "vddio", 10000 },
> > > > + { "vddxo", 80000 },
> > > > + { "vddrf", 300000 },
> > > > + { "vddch0", 450000 },
> > > > },
> > > > .num_vregs = 4,
> > > > };
> > > > @@ -1386,13 +1384,6 @@ static int qca_power_off(struct hci_dev *hdev)
> > > > static int qca_enable_regulator(struct qca_vreg vregs,
> > > > struct regulator *regulator)
> > > > {
> > > > - int ret;
> > > > -
> > > > - ret = regulator_set_voltage(regulator, vregs.min_uV,
> > > > - vregs.max_uV);
> > > > - if (ret)
> > > > - return ret;
> > > > -
> > > > return regulator_enable(regulator);
> > > >
> > > > }
> > > > @@ -1401,7 +1392,6 @@ static void qca_disable_regulator(struct
> > > > qca_vreg vregs,
> > > > struct regulator *regulator)
> > > > {
> > > > regulator_disable(regulator);
> > > > - regulator_set_voltage(regulator, 0, vregs.max_uV);
> > > >
> > > > }
> > >
> > > This was brought up multiple times during the initial review, but
> > > wasn't addressed.
> > >
> > > Reviewed-by: Matthias Kaehlcke <mka@...omium.org>
>
>
> yes true PMIC dts regulator should do this.
> But we have some real time issues observed.
>
> Issue 1:
>
> In PMIC dts node, ASAIK we have three levels of voltages.
>
> 1. Default voltage.
> 2. Minimum voltage. (mandatory entry)
> 3. Maximum voltage. (mandatory entry)
>
> Let us assume that the if PMIC regulator dts node supports defaults voltage
> to 3.2 Volts and Min as 3.1 V and max as 3.3V
> So default operating voltage is 3.1V when we turn on BT (but according to
> spec SoC requires min of 3.3V to operate,
> Might have some functionality failures on end to end testing
The PMIC regulator shouldn't be configured with the entire range of voltages
it can generate, but with a range of voltages that is suitable for all its
consumers.
In other words if BT requires a minimum voltage of 3.3V the minimum voltage
of the regulator should be at least 3.3V.
> Issue 2:
>
> WCN3990 RF is shared with WiFi, so it also try to turn on the regulators.
> Wifi driver uses the same approach of restricting to min and max voltages in
> driver.
> Let us assume we turned ON BT and CH0 is set to 3.1v (as in your case), Wifi
> is tuned on now, as its request the CH0 to operate at 3.3 Volts, regulator
> will fail to set this voltage as BT is operating
> at CH0 3.1v (assuming max voltage is 3.2v)
> https://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth-next.git/tree/drivers/net/wireless/ath/ath10k/snoc.c#n39
see above
> Issue 3:
>
> By mistake PMIC has low min or default voltage and high max voltages, that
> is harm for WNC3990.
>
> I would suggest to restrict the min and max voltages in driver, instead of
> relaying on PMIC to do this.
> BTW pmic will do this and doing it in our driver is safer.
What if another device switches the regulator on before BT?
Again, what you describe is a misconfiguration of the regulator and should
be fixed at its root, instead of implementing unreliable 'safeguards' in each
and every driver using regulators.
Powered by blists - more mailing lists