Re: [PATCH 2/2] drm/radeon/kms: wrap-up handling of floating connectors and connector unavailability status logging

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, Nov 30, 2011 at 5:54 AM, Thomas Reim <reimth@xxxxxxxxxxxxxx> wrote:
> Am Dienstag, den 29.11.2011, 15:53 -0500 schrieb Alex Deucher:
>> On Tue, Nov 29, 2011 at 1:06 PM, Thomas Reim <reimth@xxxxxxxxxxxxxx> wrote:
>> > Dear Alex,
>> >
>> >> On Mon, Nov 28, 2011 at 11:20 AM, Thomas Reim <reimth@xxxxxxxxxxxxxx> wrote:
>> >> >        Extended DDC probe is now default for RADEON chipsets. In case of
>> >> >        HW bugs (e. g. floating connectors), the affected connectors will
>> >> >        not be used, as a valid EDID header can not be detected. Another
>> >> >        patch removed DDC detection and connector status logging during
>> >> >        device setup. So the user is not informed anymore about HW bugs
>> >> >        leading to connectors being unavailable.
>> >> >        Reintroduce one-time logging of connector unavailability status
>> >> >        when probing (single) connector modes during framebuffer
>> >> >        initialisation.
>> >> >
>> >> >        For RS690 chipsets DDC detection shall be stopped, if the i2c bus
>> >> >        receives all-0 EDIDs (floating connectors). The introduction of
>> >> >        extended DDC probing prevents the driver from doing this. Correctly
>> >> >        relocate the related code.
>> >>
>> >> Is any of this needed anymore now that we do an extended probe?  I
>> >> think the original null edid patch was just papering over the actual
>> >> issue which was the need for a full edid header probe.
>> >
>> > The difference, at least this is my understanding of Dave's patch, is
>> > that we completely stop DDC detection for the famous RS690/RS740 chips
>> > with missing or disabled HDMI add-on cards. We decrease latency, as we
>> > do not retrieve data from i2c bus anymore. Therefore, I would keep it.
>> >
>>
>> My only concern is that we may disable the i2c bus if there is no
>> monitor connected which would prevent edid fetching from working in
>> the future.  Does the problematic ddia i2c line produce the same error
>> in the following cases:
>>
>
> The proposed solution differentiates between i2c bus responds, i. e. a
> DDC is available, and EDID (header) is valid.

Right, but my concern is that if we remove the i2c bus, we no longer get DDC.

>
>> 1. ddia hdmi connector present, no monitor attached
>
> DDC is not available: The connector is regarded as being disconnected;
> function drm_helper_probe_single_connector_modes() records a debug
> message in the logs, i. e. "HDMI-A-1 is disconnected".
>
> The connector will be probed again, when probing single connector modes
> (using function drm_helper_probe_single_connector_modes() about every
> second).

It's every 5 seconds for polled connectors.  If the OEM implements an
HPD pin for the connector, it's interrupt driven, so no polling
occurs.  We get an interrupt when a monitor is connected.

>
>> 2. ddia hdmi connector present, monitor with faulty edid attached
>>
>
> DDC is available, invalid EDID (header): The connector is regarded as
> being disconnected; If it's the first time, that this connector has been
> probed, usually during framebuffer initialisation, dump the faulty EDID
> to the logs and add an info message, that there is a problem with the
> monitor: e. g. "HDMI-A-1: probed a monitor, DDC responded but no|invalid
> EDID". In addition, the debug message of case 1 is logged.
>
> Then we have two subcases:
> a) General
> The connector will be probed again, when probing single connector modes
> (using function drm_helper_probe_single_connector_modes() about every
> second). But there will be only the debug message logged. No info
> message and no EDID dump.
> b) EDID is all zero, i. e. not more than 8 non-zero bytes have been
> received (RS690/RS740 chips only)
> Floating connector is assumed and an info message is logged: "HDMI-A-1:
> detected RS690 floating bus bug, stopping ddc detect". Such a connector
> will not be probed again, reducing latency. I believe that this is
> rather a theoretical case, as it requires a monitor with an erased
> EEPROM.
>
>> 3. ddia hdmi connector absent
> a) Correctly implemented HW
> would handle this in the same way as for case 1. The connector is part
> of the chipset, i2c bus is correctly terminated and will never be used,
> but being probed. This is a drawback of implmentations, that use a
> chipset wich provides more connectors than being used by the
> implementation.
> b) Buggy HW
> The connector's i2c has not been terminated correctly. It can't be used,
> but is detected. The EDID information will mainly consist of zero bytes,
> except of some random byte values.
>
> DDC is available, invalid EDID (header): The connector is regarded as
> being disconnected; If it's the first time, that this connector has been
> probed, usually during framebuffer initialisation, dump the received information
> of the not available EDID to the logs and add an info message, that
> there is a problem with the connector: e. g. "HDMI-A-1: probed a
> monitor, DDC responded but no|invalid EDID". In addition, the debug
> message of the first case is logged.
>
> Then we have two subcases:
> (1) RS690/RS740 chips
> Floating connector is assumed and an info message is logged: "HDMI-A-1:
> detected RS690 floating bus bug, stopping ddc detect". Such a connector
> will not be probed again, reducing latency.
> (2) Other chipsets
> The connector will be probed again, when probing single connector modes
> (using function drm_helper_probe_single_connector_modes() about every
> second). But there will be only a debug message (see case 1) logged. No
> info message and EDID dump.
>
>>
>> In the third case it makes sense to remove the i2c bus, but in the
>> first two, we may remove the i2c bus which would prevent future edid
>> fetches from working.
>
> The there is a 4th use case:
> 4. ddia hdmi connector present, monitor with faulty edid connected, then
> removed and re-connected again
> The Radeon driver will behave in the same way as in case 2. But each
> time the faulty monitor is attached again to the connector, the info
> message extended by a dump of the faulty EDID will be recorded once (!)
> in the logs.
>

Right.  My concern is the following.  A user boots his system with a
ddia connector present, but no monitor connected to it.  The driver
sees a faulty EDID header and removes the DDC bus.  Later the user
connects a monitor with a working EDID and tries to enable it.  Now
since the i2c bus is gone we no longer have DDC so we can't get an
EDID for the monitor and will probably not detect anything connected.
If we leave things as is, everything should just work.

>>
>> >>
>> >> >
>> >> > Signed-off-by: Thomas Reim <reimth@xxxxxxxxx>
>> >> > ---
>> >> >  drivers/gpu/drm/radeon/radeon_connectors.c |   54 +++++++++++++++++++++++-----
>> >> >  drivers/gpu/drm/radeon/radeon_i2c.c        |    7 +++-
>> >> >  drivers/gpu/drm/radeon/radeon_mode.h       |    3 ++
>> >> >  3 files changed, 53 insertions(+), 11 deletions(-)
>> >> >
>> >> > diff --git a/drivers/gpu/drm/radeon/radeon_connectors.c b/drivers/gpu/drm/radeon/radeon_connectors.c
>> >> > index e7cb3ab..a4c0f59 100644
>> >> > --- a/drivers/gpu/drm/radeon/radeon_connectors.c
>> >> > +++ b/drivers/gpu/drm/radeon/radeon_connectors.c
>> >> > @@ -662,6 +662,7 @@ radeon_vga_detect(struct drm_connector *connector, bool force)
>> >> >        struct radeon_connector *radeon_connector = to_radeon_connector(connector);
>> >> >        struct drm_encoder *encoder;
>> >> >        struct drm_encoder_helper_funcs *encoder_funcs;
>> >> > +       struct edid *edid = NULL;
>> >> >        bool dret = false;
>> >> >        enum drm_connector_status ret = connector_status_disconnected;
>> >> >
>> >> > @@ -669,8 +670,24 @@ radeon_vga_detect(struct drm_connector *connector, bool force)
>> >> >        if (!encoder)
>> >> >                ret = connector_status_disconnected;
>> >> >
>> >> > -       if (radeon_connector->ddc_bus)
>> >> > +       if (radeon_connector->ddc_bus) {
>> >> >                dret = radeon_ddc_probe(radeon_connector);
>> >> > +               if (!dret && radeon_connector->broken_edid_header_counter == 1) {
>> >> > +                       /* Found a broken DDC when probing (single) connector
>> >> > +                          modes during framebuffer initialisation.
>> >> > +                          Clean-up and log the HW bug. */
>> >> > +                       edid = drm_get_edid(&radeon_connector->base, &radeon_connector->ddc_bus->adapter);
>> >> > +                       if (!edid) {
>> >> > +                               radeon_connector->broken_edid_header_counter++;
>> >> > +                               DRM_INFO("%s: probed a monitor, DDC responded but no|invalid EDID\n",
>> >> > +                                         drm_get_connector_name(connector));
>> >> > +                       } else {
>> >> > +                               /* We should not get here, as the EDID header is broken */
>> >> > +                               connector->display_info.raw_edid = NULL;
>> >> > +                               kfree(edid);
>> >> > +                       }
>> >> > +               }
>> >> > +       }
>> >>
>> >> What is this hunk for?  It's not related to the problematic DDIA
>> >> interface which is digital only.  It seems like it will just spam the
>> >> log for VGA monitors with faulty EDIDs.
>> >
>> > The broken_edid_header_counter prevents from spamming. Only the first
>> > time, the extended DDC probe fails due to a broken EDID header (which
>> > may happen also to VGA connectors with a buggy monitor connected; at
>> > least I remember to have seen such dmesg logs in the past) we would
>> > inform users. As the EDID is broken, the screen will stay dark, but
>> > users would have at least the required information in the logs. They can
>> > post them, once they've connected a working monitor.
>> > My ASUS RS690 system suffered from such log spamming. So please believe
>> > me, that I would never ask for a solution that leads us back to this.
>> >
>>
>> I think I would skip this part for now. radeon_vga_detect() gets
>> called anytime you request a probe of a vga connector.  The full
>> header check in the radeon_ddc_probe should make this a moot point in
>> most cases.  It's also inconsistent to have this behaviour for only
>> certain connectors.
>
> For VGA connectors we focus on case 1 and 2:
> 1. no monitor attached
>
> DDC is not available: The connector is regarded as being disconnected;
> function drm_helper_probe_single_connector_modes() records a debug
> message in the logs, i. e. "VGA-1 is disconnected".

For VGA we can still use load detection.  Plus we now introduce a full
EDID fetch.

>
> The connector will be probed again, when probing single connector modes
> (using function drm_helper_probe_single_connector_modes() about every
> second).
>
> 2. monitor with faulty edid attached
>
> DDC is available, invalid EDID (header): The connector is regarded as
> being disconnected; If it's the first time, that this connector has been
> probed, usually during framebuffer initialisation, dump the faulty EDID
> to the logs and add an info message, that there is a problem with the
> monitor: e. g. "VGA-1: probed a monitor, DDC responded but no|invalid
> EDID". In addition, the debug message of case 1 is logged.
>
> The connector will be probed again, when probing single connector modes
> (using function drm_helper_probe_single_connector_modes() about every
> second). But there will be only the debug message logged. No info
> message and no EDID dump.
>
> We could skip this part of the patch. But if we did this, this scenario
> (http://askubuntu.com/questions/14299/my-monitor-is-plugged-into-vga-0-why-it-is-giving-me-errors-about-vga-1) would lead to a dark screen with some meaningless debug only information in the dmesg log:
> ...
> [drm:drm_helper_probe_single_connector_modes], VGA-1
> [drm:drm_helper_probe_single_connector_modes], VGA-1 is disconnected
> ...
>

With VGA load detection will mostly likely detect a monitor even if
and EDID is not available.

> As a non-expert user I would prefer a more informative log, that helps
> me to request support:

If we want more complex detection logic I'd prefer we do it in the
common drm code so it's consistent across cards and connectors.  Also,
I think we should be more forgiving about wonky EDIDs but that's
another issue.

> ...
> [drm:drm_helper_probe_single_connector_modes], VGA-1
> [drm:drm_edid_block_valid] *ERROR* EDID checksum is invalid, remainder
> is 126
> [drm:drm_edid_block_valid] *ERROR* Raw EDID:
> <3>ee 8b c5 a4 5a 48 9b 25 1a 50 54 bf ef 00 81 80  ....ZH.%.PT.....
> <3>10 10 01 03 68 22 1b 78 ee 8b c5 a4 5a 48 9b 25  ....h".x....ZH.%
> <3>1a 50 54 bf ef 00 81 80 71 4f 01 01 01 01 01 01  .PT.....qO......
> <3>01 01 01 01 01 01 30 2a 00 98 51 00 2a 40 30 70  ......0*..Q.*@0p
> <3>13 00 52 0e 11 00 00 1e 00 00 00 fd 00 38 4c 1e  ..R..........8L.
> <3>52 0e 00 0a 20 20 20 20 20 20 00 00 00 fc 00 43  R...      .....C
> <3>4d 43 20 31 37 0a 20 20 20 20 20 20 00 00 00 ff  MC 17.      ....
> <3>00 30 0a 20 20 20 20 20 20 20 20 20 20 20 00 9d  .0.           ..
>
> [drm:drm_edid_block_valid] *ERROR* EDID checksum is invalid, remainder is 126
> [drm:drm_edid_block_valid] *ERROR* Raw EDID:
> <3>ee 8b c5 a4 5a 48 9b 25 1a 50 54 bf ef 00 81 80  ....ZH.%.PT.....
> <3>10 10 01 03 68 22 1b 78 ee 8b c5 a4 5a 48 9b 25  ....h".x....ZH.%
> <3>1a 50 54 bf ef 00 81 80 71 4f 01 01 01 01 01 01  .PT.....qO......
> <3>01 01 01 01 01 01 30 2a 00 98 51 00 2a 40 30 70  ......0*..Q.*@0p
> <3>13 00 52 0e 11 00 00 1e 00 00 00 fd 00 38 4c 1e  ..R..........8L.
> <3>52 0e 00 0a 20 20 20 20 20 20 00 00 00 fc 00 43  R...      .....C
> <3>4d 43 20 31 37 0a 20 20 20 20 20 20 00 00 00 ff  MC 17.      ....
> <3>00 30 0a 20 20 20 20 20 20 20 20 20 20 20 00 9d  .0.           ..
>
> [drm:drm_edid_block_valid] *ERROR* EDID checksum is invalid, remainder is 126
> [drm:drm_edid_block_valid] *ERROR* Raw EDID:
> <3>ee 8b c5 a4 5a 48 9b 25 1a 50 54 bf ef 00 81 80  ....ZH.%.PT.....
> <3>10 10 01 03 68 22 1b 78 ee 8b c5 a4 5a 48 9b 25  ....h".x....ZH.%
> <3>1a 50 54 bf ef 00 81 80 71 4f 01 01 01 01 01 01  .PT.....qO......
> <3>01 01 01 01 01 01 30 2a 00 98 51 00 2a 40 30 70  ......0*..Q.*@0p
> <3>13 00 52 0e 11 00 00 1e 00 00 00 fd 00 38 4c 1e  ..R..........8L.
> <3>52 0e 00 0a 20 20 20 20 20 20 00 00 00 fc 00 43  R...      .....C
> <3>4d 43 20 31 37 0a 20 20 20 20 20 20 00 00 00 ff  MC 17.      ....
> <3>00 30 0a 20 20 20 20 20 20 20 20 20 20 20 00 9d  .0.           ..
>
> [drm:drm_edid_block_valid] *ERROR* EDID checksum is invalid, remainder is 126
> [drm:drm_edid_block_valid] *ERROR* Raw EDID:
> <3>ee 8b c5 a4 5a 48 9b 25 1a 50 54 bf ef 00 81 80  ....ZH.%.PT.....
> <3>10 10 01 03 68 22 1b 78 ee 8b c5 a4 5a 48 9b 25  ....h".x....ZH.%
> <3>1a 50 54 bf ef 00 81 80 71 4f 01 01 01 01 01 01  .PT.....qO......
> <3>01 01 01 01 01 01 30 2a 00 98 51 00 2a 40 30 70  ......0*..Q.*@0p
> <3>13 00 52 0e 11 00 00 1e 00 00 00 fd 00 38 4c 1e  ..R..........8L.
> <3>52 0e 00 0a 20 20 20 20 20 20 00 00 00 fc 00 43  R...      .....C
> <3>4d 43 20 31 37 0a 20 20 20 20 20 20 00 00 00 ff  MC 17.      ....
> <3>00 30 0a 20 20 20 20 20 20 20 20 20 20 20 00 9d  .0.           ..
>
> radeon 0000:01:05.0: VGA-1: EDID block 0 invalid.
> [drm:radeon_vga_detect] VGA-1: probed a monitor, DDC responded but no|invalid EDID
> [drm:drm_helper_probe_single_connector_modes], VGA-1 is disconnected
> ...
>
> We see, that the EDID information is stable, but wrong, i. e. no EDID
> header. The monitor seems to be defect. Therefore, the system does not
> use this connector/monitor.

The user just sees a blank screen whether there is a bunch of EDID
junk or not doesn't really change things.  It's not like we can fix
their monitor.

>
> Furtheron, only the followuing two debug messages will be in the log
> (about every second). This is also true, if the user disconnects the
> buggy monitor:
> ...
> [drm:drm_helper_probe_single_connector_modes], VGA-1
> [drm:drm_helper_probe_single_connector_modes], VGA-1 is disconnected
> ...
> [drm:drm_helper_probe_single_connector_modes], VGA-1
> [drm:drm_helper_probe_single_connector_modes], VGA-1 is disconnected
> ...
> [drm:drm_helper_probe_single_connector_modes], VGA-1
> [drm:drm_helper_probe_single_connector_modes], VGA-1 is disconnected
> ...
> If the debug option is disabled, no more information will be recorded in
> the logs.
>
> If the user attaches a working monitor, the screen will be back and
> following debug messages will be in the logs (every second):
> ...
> [drm:drm_helper_probe_single_connector_modes], VGA-1
> [drm:drm_helper_probe_single_connector_modes], Probed modes for VGA-1
> [drm:drm_mode_debug_printmodeline], Modeline 17:"1280x1024" 60 108000
> 1280 1328 1440 1688 1024 1025 1028 1066 0x48 0x5
> ...
> [drm:drm_helper_probe_single_connector_modes], VGA-1
> [drm:drm_helper_probe_single_connector_modes], Probed modes for VGA-1
> [drm:drm_mode_debug_printmodeline], Modeline 17:"1280x1024" 60 108000
> 1280 1328 1440 1688 1024 1025 1028 1066 0x48 0x5
> ...
> [drm:drm_helper_probe_single_connector_modes], VGA-1
> [drm:drm_helper_probe_single_connector_modes], Probed modes for VGA-1
> [drm:drm_mode_debug_printmodeline], Modeline 17:"1280x1024" 60 108000
> 1280 1328 1440 1688 1024 1025 1028 1066 0x48 0x5
> ...
>
> This behaviour of the Radeon driver is consistent for VGA and DVI
> connectors. Current LVDS implementation will still flood the logs with
> EDID dumps in case of faulty EDIDs, as there is no DDC probe
> implemented. In case of TV connectors, we don't have EDIDs. Only
> exception are DP connectors, where we still have to rely on the
> drm_helper_probe_single_connector_modes() debug log messages.
>
>>
>> >>
>> >> >        if (dret) {
>> >> >                radeon_connector->detected_by_load = false;
>> >> >                if (radeon_connector->edid) {
>> >> > @@ -847,12 +864,38 @@ radeon_dvi_detect(struct drm_connector *connector, bool force)
>> >> >        struct drm_encoder *encoder = NULL;
>> >> >        struct drm_encoder_helper_funcs *encoder_funcs;
>> >> >        struct drm_mode_object *obj;
>> >> > +       struct edid *edid = NULL;
>> >> >        int i;
>> >> >        enum drm_connector_status ret = connector_status_disconnected;
>> >> >        bool dret = false;
>> >> >
>> >> > -       if (radeon_connector->ddc_bus)
>> >> > +       if (radeon_connector->ddc_bus) {
>> >> >                dret = radeon_ddc_probe(radeon_connector);
>> >> > +               if (!dret && radeon_connector->broken_edid_header_counter == 1) {
>> >> > +                       /* Found a broken DDC when probing (single) connector
>> >> > +                          modes during framebuffer initialisation.
>> >> > +                          Clean-up and log the HW bug. */
>> >> > +                       edid = drm_get_edid(&radeon_connector->base, &radeon_connector->ddc_bus->adapter);
>> >> > +                       if (!edid) {
>> >> > +                               radeon_connector->broken_edid_header_counter++;
>> >> > +                               DRM_INFO("%s: probed a monitor, DDC responded but no|invalid EDID\n",
>> >> > +                                         drm_get_connector_name(connector));
>> >> > +                               /* rs690 seems to have a problem with connectors not existing and return
>> >> > +                                * random EDIDs mainly with blocks of 0's. If we see this, just stop
>> >> > +                                * polling on this output */
>> >> > +                               if ((rdev->family == CHIP_RS690 || rdev->family == CHIP_RS740)) {
>> >> > +                                       ret = connector_status_disconnected;
>> >> > +                                       DRM_INFO("%s: detected RS690 floating bus bug, stopping ddc detect\n",
>> >> > +                                                 drm_get_connector_name(connector));
>> >> > +                                       radeon_connector->ddc_bus = NULL;
>> >> > +                               }
>> >> > +                       } else {
>> >> > +                               /* We should not get here, as the EDID header is broken */
>> >> > +                               connector->display_info.raw_edid = NULL;
>> >> > +                               kfree(edid);
>> >> > +                       }
>> >> > +               }
>> >> > +       }
>> >>
>> >> Is this really needed?  What is it preventing?  Doesn't the full EDID
>> >> header probe alleviate the need for this?  My concern is that we will
>> >> remove the ddc bus on connectors that are valid just because the user
>> >> probed with no monitor connected or a monitor with a faulty edid.
>> >> Then all future probes will fail to get an EDID since the bus is gone.
>> >
>> > You're right. This is a bug! Somehow I lost the third if condition: &&
>> > radeon_connector->base.null_edid_counter when updating the patches.
>> > Thanks for checking the code. I'll post the corrected one after sending
>> > this mail.
>> >
>> >>
>> >> >        if (dret) {
>> >> >                radeon_connector->detected_by_load = false;
>> >> >                if (radeon_connector->edid) {
>> >> > @@ -864,13 +907,6 @@ radeon_dvi_detect(struct drm_connector *connector, bool force)
>> >> >                if (!radeon_connector->edid) {
>> >> >                        DRM_ERROR("%s: probed a monitor but no|invalid EDID\n",
>> >> >                                        drm_get_connector_name(connector));
>> >> > -                       /* rs690 seems to have a problem with connectors not existing and always
>> >> > -                        * return a block of 0's. If we see this just stop polling on this output */
>> >> > -                       if ((rdev->family == CHIP_RS690 || rdev->family == CHIP_RS740) && radeon_connector->base.null_edid_counter) {
>> >> > -                               ret = connector_status_disconnected;
>> >> > -                               DRM_ERROR("%s: detected RS690 floating bus bug, stopping ddc detect\n", drm_get_connector_name(connector));
>> >> > -                               radeon_connector->ddc_bus = NULL;
>> >> > -                       }
>> >>
>> >> This part can probably be removed now.
>> >>
>> >> >                } else {
>> >> >                        radeon_connector->use_digital = !!(radeon_connector->edid->input & DRM_EDID_INPUT_DIGITAL);
>> >> >
>> >> > diff --git a/drivers/gpu/drm/radeon/radeon_i2c.c b/drivers/gpu/drm/radeon/radeon_i2c.c
>> >> > index 7bb1b07..ac6aa02 100644
>> >> > --- a/drivers/gpu/drm/radeon/radeon_i2c.c
>> >> > +++ b/drivers/gpu/drm/radeon/radeon_i2c.c
>> >> > @@ -59,10 +59,12 @@ bool radeon_ddc_probe(struct radeon_connector *radeon_connector)
>> >> >                radeon_router_select_ddc_port(radeon_connector);
>> >> >
>> >> >        ret = i2c_transfer(&radeon_connector->ddc_bus->adapter, msgs, 2);
>> >> > -       if (ret != 2)
>> >> > +       if (ret != 2) {
>> >> >                /* Couldn't find an accessible DDC on this connector */
>> >> > +               radeon_connector->broken_edid_header_counter = 0;
>> >> >                return false;
>> >> > -       /* Probe also for valid EDID header
>> >> > +       }
>> >> > +       /* DDC is available; probe also for valid EDID header
>> >> >         * EDID header starts with:
>> >> >         * 0x00,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0x00.
>> >> >         * Only the first 6 bytes must be valid as
>> >> > @@ -70,6 +72,7 @@ bool radeon_ddc_probe(struct radeon_connector *radeon_connector)
>> >> >        if (drm_edid_header_is_valid(buf) < 6) {
>> >> >                /* Couldn't find an accessible EDID on this
>> >> >                 * connector */
>> >> > +               radeon_connector->broken_edid_header_counter++;
>> >> >                return false;
>> >> >        }
>> >> >        return true;
>> >> > diff --git a/drivers/gpu/drm/radeon/radeon_mode.h b/drivers/gpu/drm/radeon/radeon_mode.h
>> >> > index 2c2e75e..4af4583 100644
>> >> > --- a/drivers/gpu/drm/radeon/radeon_mode.h
>> >> > +++ b/drivers/gpu/drm/radeon/radeon_mode.h
>> >> > @@ -442,6 +442,9 @@ struct radeon_connector {
>> >> >        /* we need to mind the EDID between detect
>> >> >           and get modes due to analog/digital/tvencoder */
>> >> >        struct edid *edid;
>> >> > +       /* extended DDC probing is now default for radeon connectors
>> >> > +          count the number of failed EDID header probes */
>> >> > +       int broken_edid_header_counter;
>> >> >        void *con_priv;
>> >> >        bool dac_load_detect;
>> >> >        bool detected_by_load; /* if the connection status was determined by load */
>> >> > --
>> >> > 1.7.1
>> >> >
>> >> >
>> >
>> > Best regards
>> >
>> > Thomas
>> >
>> >
>> >
>
>
>
_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
http://lists.freedesktop.org/mailman/listinfo/dri-devel



[Index of Archives]     [Linux DRI Users]     [Linux Intel Graphics]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]
  Powered by Linux