add EGL_EXT_device_query_pci

[1ace]

Firefox requested to have a way to query the device's vendor PCI ID and device PCI ID, so here's an extension that adds this.

Cross-references:

• spec (here)
• libglvnd ext addition
• Mesa implementation
• implementation test

/cc @rmader

[kbrenneman]

Would it make sense to add queries for the domain/bus/slot ID's as well?

[1ace]

Would it make sense to add queries for the domain/bus/slot ID's as well?

Why not, but is there a use-case for these?

[kbrenneman]

One use I can think of would be to distinguish between multiple identical devices.

[1ace]

It way beyond the problem we're trying to solve here, but it's trivial to implement and conceptually it belongs in the same extension so I'm adding these and re-renaming the spec to EGL_EXT_device_query_pci 👍

[kbrenneman]

As a minor nitpick, the "The query always succeeds and returns EGL_TRUE, and is set appropriately" line probably isn't necessary, since that's the default assumption for EGL functions.

Other than generic errors (invalid EGLDeviceEXT handle, out of memory, etc), EGL functions list the errors that they can produce.

In this case, it doesn't make much difference, though, since I don't think an implementation could run into any internal errors for those queries. Anything that did go wrong would have gone wrong in eglQueryDevicesEXT instead.

Anyway, change that if you want. The spec looks good to me either way.

[1ace]

Thanks for the feedback; you're right, I'll drop the line.

I also was wondering if I should explicitly say that EGL_QUERY_PCI_VENDOR_ID_EXT is the vendor ID, etc. or if it's all obvious enough as it is?

[kbrenneman]

I'd say it's obvious enough what each attribute is.

[stonesthrow]

Change Milestone to "Approved to Merge" when ready and assign to oddhack for Jon to merge.

[1ace]

@stonesthrow I don't have permission to change the milestones; could you give me permission, or do it for me? Thanks!

[stonesthrow]

@oddhack: Ready to merge

[1ace]

Actually can this wait? @evelikov has requested a change of enum values, I'm waiting on @fooishbar to confirm.

[oddhack]

The extension number here and in the spec text will need to be updated when the PR is actually accepted, up through 141 are used at the moment.

[oddhack]

@1ace @evelikov @fooishbar where does this extension stand - was your issue about enum values resolved? Would be nice to get it finished and published.