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]
Date:   Tue, 15 Feb 2022 14:33:38 +0200
From:   Heikki Krogerus <heikki.krogerus@...ux.intel.com>
To:     Jack Pham <quic_jackp@...cinc.com>
Cc:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Guenter Roeck <linux@...ck-us.net>,
        Benson Leung <bleung@...gle.com>,
        Prashant Malani <pmalani@...omium.org>,
        Jameson Thies <jthies@...gle.com>,
        "Regupathy, Rajaram" <rajaram.regupathy@...el.com>,
        linux-usb@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v1 3/3] usb: typec: ucsi: Register USB Power Delivery
 Capabilities

Hi,

On Wed, Feb 09, 2022 at 11:56:11PM -0800, Jack Pham wrote:
> Hi Heikki,
> 
> On Thu, Feb 03, 2022 at 05:46:57PM +0300, Heikki Krogerus wrote:
> > UCSI allows the USB PD capabilities to be read with the
> > GET_PDO command. This will register those capabilities, and
> > that way make them visible to the user space.
> > 
> > Signed-off-by: Heikki Krogerus <heikki.krogerus@...ux.intel.com>
> > ---
> >  drivers/usb/typec/ucsi/ucsi.c | 128 +++++++++++++++++++++++++++++++---
> >  drivers/usb/typec/ucsi/ucsi.h |   8 +++
> >  2 files changed, 125 insertions(+), 11 deletions(-)
> > 
> > diff --git a/drivers/usb/typec/ucsi/ucsi.c b/drivers/usb/typec/ucsi/ucsi.c
> > index f0c2fa19f3e0f..5149001093c7f 100644
> > --- a/drivers/usb/typec/ucsi/ucsi.c
> > +++ b/drivers/usb/typec/ucsi/ucsi.c
> > @@ -568,8 +568,8 @@ static void ucsi_unregister_altmodes(struct ucsi_connector *con, u8 recipient)
> >  	}
> >  }
> >  
> > -static int ucsi_get_pdos(struct ucsi_connector *con, int is_partner,
> > -			 u32 *pdos, int offset, int num_pdos)
> > +static int ucsi_read_pdos(struct ucsi_connector *con, enum typec_role role, int is_partner,
> > +			  u32 *pdos, int offset, int num_pdos)
> >  {
> >  	struct ucsi *ucsi = con->ucsi;
> >  	u64 command;
> > @@ -579,7 +579,7 @@ static int ucsi_get_pdos(struct ucsi_connector *con, int is_partner,
> >  	command |= UCSI_GET_PDOS_PARTNER_PDO(is_partner);
> >  	command |= UCSI_GET_PDOS_PDO_OFFSET(offset);
> >  	command |= UCSI_GET_PDOS_NUM_PDOS(num_pdos - 1);
> > -	command |= UCSI_GET_PDOS_SRC_PDOS;
> > +	command |= is_source(role) ? UCSI_GET_PDOS_SRC_PDOS : 0;
> >  	ret = ucsi_send_command(ucsi, command, pdos + offset,
> >  				num_pdos * sizeof(u32));
> >  	if (ret < 0 && ret != -ETIMEDOUT)
> > @@ -590,26 +590,39 @@ static int ucsi_get_pdos(struct ucsi_connector *con, int is_partner,
> >  	return ret;
> >  }
> >  
> > -static int ucsi_get_src_pdos(struct ucsi_connector *con)
> > +static int ucsi_get_pdos(struct ucsi_connector *con, enum typec_role role,
> > +			 int is_partner, u32 *pdos)
> >  {
> > +	u8 num_pdos;
> >  	int ret;
> >  
> >  	/* UCSI max payload means only getting at most 4 PDOs at a time */
> > -	ret = ucsi_get_pdos(con, 1, con->src_pdos, 0, UCSI_MAX_PDOS);
> > +	ret = ucsi_read_pdos(con, role, is_partner, pdos, 0, UCSI_MAX_PDOS);
> >  	if (ret < 0)
> >  		return ret;
> >  
> > -	con->num_pdos = ret / sizeof(u32); /* number of bytes to 32-bit PDOs */
> > -	if (con->num_pdos < UCSI_MAX_PDOS)
> > -		return 0;
> > +	num_pdos = ret / sizeof(u32); /* number of bytes to 32-bit PDOs */
> > +	if (num_pdos < UCSI_MAX_PDOS)
> > +		return num_pdos;
> >  
> >  	/* get the remaining PDOs, if any */
> > -	ret = ucsi_get_pdos(con, 1, con->src_pdos, UCSI_MAX_PDOS,
> > -			    PDO_MAX_OBJECTS - UCSI_MAX_PDOS);
> > +	ret = ucsi_read_pdos(con, role, is_partner, pdos, UCSI_MAX_PDOS,
> > +			     PDO_MAX_OBJECTS - UCSI_MAX_PDOS);
> > +	if (ret < 0)
> > +		return ret;
> > +
> > +	return ret / sizeof(u32) + num_pdos;
> > +}
> > +
> > +static int ucsi_get_src_pdos(struct ucsi_connector *con)
> > +{
> > +	int ret;
> > +
> > +	ret = ucsi_get_pdos(con, TYPEC_SOURCE, 1, con->src_pdos);
> 
> This issues the GET_PDOS command to PPM to get the source PDOs of the
> partner...
> 
> >  	if (ret < 0)
> >  		return ret;
> >  
> > -	con->num_pdos += ret / sizeof(u32);
> > +	con->num_pdos += ret;
> >  
> >  	ucsi_port_psy_changed(con);
> >  
> > @@ -638,6 +651,60 @@ static int ucsi_check_altmodes(struct ucsi_connector *con)
> >  	return ret;
> >  }
> >  
> > +static int ucsi_register_partner_pdos(struct ucsi_connector *con)
> > +{
> > +	struct pd_desc desc = { con->ucsi->cap.pd_version };
> > +	struct pd_capabilities *cap;
> > +	struct pd_caps_desc caps;
> > +	int ret;
> > +
> > +	con->partner_pd = typec_partner_register_pd(con->partner, &desc);
> > +	if (IS_ERR(con->partner_pd))
> > +		return PTR_ERR(con->partner_pd);
> > +
> > +	ret = ucsi_get_pdos(con, TYPEC_SOURCE, 1, caps.pdo);
> 
> ... and also here.
> 
> > +	if (ret < 0)
> > +		return ret;
> > +
> > +	if (ret < PDO_MAX_OBJECTS)
> > +		caps.pdo[ret] = 0;
> > +	caps.role = TYPEC_SOURCE;
> > +
> > +	cap = pd_register_capabilities(con->partner_pd, &caps);
> > +	if (IS_ERR(cap))
> > +		return PTR_ERR(cap);
> > +
> > +	ret = typec_partner_set_pd_capabilities(con->partner, cap);
> > +	if (ret) {
> > +		pd_unregister_capabilities(cap);
> > +		return ret;
> > +	}
> > +
> > +	con->partner_source_caps = cap;
> > +
> > +	ret = ucsi_get_pdos(con, TYPEC_SINK, 1, caps.pdo);
> > +	if (ret <= 0)
> > +		return ret;
> > +
> > +	if (ret < PDO_MAX_OBJECTS)
> > +		caps.pdo[ret] = 0;
> > +	caps.role = TYPEC_SINK;
> > +
> > +	cap = pd_register_capabilities(con->partner_pd, &caps);
> > +	if (IS_ERR(cap))
> > +		return PTR_ERR(cap);
> > +
> > +	ret = typec_partner_set_pd_capabilities(con->partner, cap);
> > +	if (ret) {
> > +		pd_unregister_capabilities(cap);
> > +		return ret;
> > +	}
> > +
> > +	con->partner_sink_caps = cap;
> > +
> > +	return 0;
> > +}
> > +
> >  static void ucsi_pwr_opmode_change(struct ucsi_connector *con)
> >  {
> >  	switch (UCSI_CONSTAT_PWR_OPMODE(con->status.flags)) {
> > @@ -646,6 +713,7 @@ static void ucsi_pwr_opmode_change(struct ucsi_connector *con)
> >  		typec_set_pwr_opmode(con->port, TYPEC_PWR_MODE_PD);
> >  		ucsi_partner_task(con, ucsi_get_src_pdos, 30, 0);
> >  		ucsi_partner_task(con, ucsi_check_altmodes, 30, 0);
> > +		ucsi_partner_task(con, ucsi_register_partner_pdos, 1, HZ);
> 
> And, both ucsi_get_src_pdos() and ucsi_register_partner_pdos() are
> scheduled to run here...
> 
> >  		break;
> >  	case UCSI_CONSTAT_PWR_OPMODE_TYPEC1_5:
> >  		con->rdo = 0;
> > @@ -704,6 +772,17 @@ static void ucsi_unregister_partner(struct ucsi_connector *con)
> >  	if (!con->partner)
> >  		return;
> >  
> > +	typec_partner_unset_pd_capabilities(con->partner, TYPEC_SINK);
> > +	pd_unregister_capabilities(con->partner_sink_caps);
> > +	con->partner_sink_caps = NULL;
> > +
> > +	typec_partner_unset_pd_capabilities(con->partner, TYPEC_SOURCE);
> > +	pd_unregister_capabilities(con->partner_source_caps);
> > +	con->partner_source_caps = NULL;
> > +
> > +	typec_partner_unregister_pd(con->partner);
> > +	con->partner_pd = NULL;
> > +
> >  	ucsi_unregister_altmodes(con, UCSI_RECIPIENT_SOP);
> >  	typec_unregister_partner(con->partner);
> >  	con->partner = NULL;
> > @@ -1037,6 +1116,8 @@ static int ucsi_register_port(struct ucsi *ucsi, int index)
> >  	u64 command;
> >  	char *name;
> >  	int ret;
> > +	struct pd_desc desc = { ucsi->cap.pd_version };
> > +	struct pd_caps_desc caps;
> >  
> >  	name = kasprintf(GFP_KERNEL, "%s-con%d", dev_name(ucsi->dev), con->num);
> >  	if (!name)
> > @@ -1103,6 +1184,24 @@ static int ucsi_register_port(struct ucsi *ucsi, int index)
> >  		goto out;
> >  	}
> >  
> > +	con->pd = typec_port_register_pd(con->port, &desc);
> > +
> > +	ret = ucsi_get_pdos(con, TYPEC_SOURCE, 0, caps.pdo);
> > +	if (ret > 0) {
> > +		caps.pdo[ret] = 0;
> > +		caps.role = TYPEC_SOURCE;
> > +		con->source_caps = pd_register_capabilities(con->pd, &caps);
> > +		typec_port_set_pd_capabilities(con->port, con->source_caps);
> > +	}
> > +
> > +	ret = ucsi_get_pdos(con, TYPEC_SINK, 0, caps.pdo);
> > +	if (ret > 0) {
> > +		caps.pdo[ret] = 0;
> > +		caps.role = TYPEC_SINK;
> > +		con->sink_caps = pd_register_capabilities(con->pd, &caps);
> > +		typec_port_set_pd_capabilities(con->port, con->sink_caps);
> > +	}
> > +
> >  	/* Alternate modes */
> >  	ret = ucsi_register_altmodes(con, UCSI_RECIPIENT_CON);
> >  	if (ret) {
> > @@ -1169,6 +1268,7 @@ static int ucsi_register_port(struct ucsi *ucsi, int index)
> >  	    UCSI_CONSTAT_PWR_OPMODE_PD) {
> >  		ucsi_get_src_pdos(con);
> >  		ucsi_check_altmodes(con);
> > +		ucsi_register_partner_pdos(con);
> 
> ... as well as here.
> 
> So wouldn't this result in the PPM issuing the same PD Get_Source_Cap
> message twice to the port partner (in either case of initial port
> registration or op mode change)?  Could we just consolidate them to just
> issue GET_PDOS only once and take care of populating the partner's
> Source Caps for both the pd_capabilties as well as power_supply purposes
> from a single helper?

Yes. I tried create as little impact on existing code with this as
possible, so I just duplicated that part, but I probable should not
have done that.

> Another aside, thinking back to a previous patch [1] I had proposed a
> few months ago, another question I had is whether it is proper to even
> issue a Get_Source_Cap message to sink-only devices, as we did encounter
> certain DisplayPort adapters that don't like when that happens.
> Wondering if it could be possible to limit calling the GET_PDOS command
> unless we know the partner is capable of operating in that particular
> power role.  e.g. don't call get_src_pdos() if partner is sink-only.
> Or is this the kind of thing that the PPM is supposed to be able to
> figure out and allow OPM to naively issue the command regardless, and
> just get an error/empty return?

Well, in many cases we simply can not rely on the PPM, but I think we
should be able rely on it with the role.

What you are proposing probable would make sense. Do you want to work
on those improvements? I don't have to modify this driver now with
this series. I also can first register the sysfs attributes from
tcpm.c.

thanks,

-- 
heikki

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ