ctlEnumerateDisplayOutputs returns CTL_RESULT_ERROR_INVALID_SIZE

[c3martin]

The first call to ctlEnumerateDisplayOutputs returns 3 display outputs, however sometimes the second call to ctlEnumerateDisplayOutputs with pCount set to 3 returns CTL_RESULT_ERROR_INVALID_SIZE.

See example below:

function main( ... )
# Get the adapter handles in hDevices as shown in the IGCL API programming guide
for(index = 0 .. Adapter_count-1)
     ctl_display_output_handle_t *hDisplayOutput = nullptr;
     # Enumerate all displays
        uint32_t Display_count = 0;
        result = ctlEnumerateDisplayOutputs(hDevices[Index], &Display_count, hDisplayOutput);
        // Display_count = 3
        // Result = CTL_RESULT_SUCCESS

        hDisplayOutput = (ctl_display_output_handle_t *)malloc(sizeof(ctl_display_output_handle_t) * Display_count);
        result = ctlEnumerateDisplayOutputs(hDevices[Index], &Display_count, hDisplayOutput);
        // Display_count = 3
        // Result = CTL_RESULT_ERROR_INVALID_SIZE

Given the above, my questions are:

1. In what circumstance would ctlEnumerateDisplayOutputs return CTL_RESULT_ERROR_INVALID_SIZE?
2. Why would ctlEnumerateDisplayOutputs tell me that there are three display outputs and then fail on the second call with CTL_RESULT_ERROR_INVALID_SIZE?

[rahulrgandhi99]

The first call to ctlEnumerateDisplayOutputs returns 3 display outputs, however sometimes the second call to ctlEnumerateDisplayOutputs with pCount set to 3 returns CTL_RESULT_ERROR_INVALID_SIZE.

See example below:

function main( ... )
# Get the adapter handles in hDevices as shown in the IGCL API programming guide
for(index = 0 .. Adapter_count-1)
     ctl_display_output_handle_t *hDisplayOutput = nullptr;
     # Enumerate all displays
        uint32_t Display_count = 0;
        result = ctlEnumerateDisplayOutputs(hDevices[Index], &Display_count, hDisplayOutput);
        // Display_count = 3
        // Result = CTL_RESULT_SUCCESS

        hDisplayOutput = (ctl_display_output_handle_t *)malloc(sizeof(ctl_display_output_handle_t) * Display_count);
        result = ctlEnumerateDisplayOutputs(hDevices[Index], &Display_count, hDisplayOutput);
        // Display_count = 3
        // Result = CTL_RESULT_ERROR_INVALID_SIZE

Given the above, my questions are:

1. In what circumstance would ctlEnumerateDisplayOutputs return CTL_RESULT_ERROR_INVALID_SIZE?
   --> This error typically happens when you pass wrong input size.
2. Why would ctlEnumerateDisplayOutputs tell me that there are three display outputs and then fail on the second call with CTL_RESULT_ERROR_INVALID_SIZE?
   --> What is your use case? Are you in b/w connecting or dis-connecting monitors.

[ashutosh-d-shukla]

@c3martin any updates on the above issue?

[c3martin]

@ashutosh-d-shukla @rahulrgandhi99

Here are some more details on the issue:

Description:
The second call to ctlEnumerateDisplayOutputs fails when the number of displays changes during the application lifetime.

To demonstrate the issue, here is a zip containing a slightly modified version of the I2C_AUX_Sample_App:

intel_test_app.zip

Reproduction Steps:

1. Ensure there is only 1 monitor connected to the target machine
2. Run the application as a Windows Service (using a tool like https://nssm.cc/)
3. Connect additional display(s) while the application is still running

Note. This issue only occurs when the application is run as a Windows Service, running as a user application does not reproduce this issue.

Logs:
// Single Display (Laptop Screen)

...
6) Scanning for Displays...
Adapter 0) Display Count: 1

// Connect 2 additional displays (3 displays in total)

7) Scanning for Displays...
Adapter 0) Display Count: 3
ctlEnumerateDisplayOutputs returned failure code: 0x4000000F
8) Scanning for Displays...
Adapter 0) Display Count: 3
ctlEnumerateDisplayOutputs returned failure code: 0x4000000F

[c3martin]

@ashutosh-d-shukla @rahulrgandhi99 are there any updates on this issue?

[ashutosh-d-shukla]

We are trying to replicate internally for this and will get back with the update.

[ashutosh-d-shukla]

Tagging @murali-i as FYI, this could be fixed in one of the latest drivers as we found a memory corruption in driver that could be causing enumDisplayOutputs failure. Will post the driver version once we have it.

[kpsam2000Intel]

By the way, @c3martin , if new displays get connected or existing ones get disconnected, it's always good to restart with a fresh query (i.e., provide NULL to buffer, get new size, provide new buffer with updated size etc.)

[ashutosh-d-shukla]

This is expected to be fixed in V172 which is yet to be released externally in drivers at intel.com.

[c3martin]

This is expected to be fixed in V172 which is yet to be released externally in drivers at intel.com.

Thank you for the update @ashutosh-d-shukla

[ashutosh-d-shukla]

@c3martin I am closing this as public drivers are released with this version. Please test with 31.0.101.5333 version from intel.com. Please reopen if you still face the problem.