Skip to contentSkip to conversation

Vdev map issues.

DISCUSSION Ben Greear 2 participants 4 messages Apr 30, 2014 Qualcomm Atheros ATH10K wireless driver

lore
Share & tools
lore

Ben Greear

author
#0 lore raw

Possibly I've made too many local changes and gotten confused, but this code appears like it should be doing an |= instead of &= in ath10k_add_interface?

err_vdev_delete:

	ath10k_wmi_vdev_delete(ar, arvif->vdev_id);
	ar->free_vdev_map &= ~(1LL << arvif->vdev_id);
	list_del(&arvif->list);

Thanks,
Ben

Signature · 8 lines
Ben Greear <[email protected]>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k
#1 lore raw
Quoted 7 lines · Ben Greear

Possibly I've made too many local changes and gotten confused, but this code appears like it should be doing an |= instead of &= in ath10k_add_interface?

err_vdev_delete:

        ath10k_wmi_vdev_delete(ar, arvif->vdev_id);
        ar->free_vdev_map &= ~(1LL << arvif->vdev_id);
        list_del(&arvif->list);

Correct. Will you submit a patch?

Michał

_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k

author
#2 lore raw
Quoted 10 lines · Michal Kazior
Quoted 7 lines · Ben Greear

Possibly I've made too many local changes and gotten confused, but this code appears like it should be doing an |= instead of &= in ath10k_add_interface?

err_vdev_delete:

        ath10k_wmi_vdev_delete(ar, arvif->vdev_id);
        ar->free_vdev_map &= ~(1LL << arvif->vdev_id);
        list_del(&arvif->list);

Correct. Will you submit a patch?

This is tangled in another patch that enables 64-bit vdev-maps, but I will rebase and post it shortly.

Another question in the related code: For monitor devices, we are using bit - 1 for the monitor device ID, but we do not subtract one when creating normal stations. Is this on purpose?

And, instead of checking for bit == 0, why not check for zero map before we even try the ffs call? That should work for __ffs64 as well, which is what I need for lots of vdevs...

Thanks,
Ben

Signature · 8 lines
Ben Greear <[email protected]>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k
author
#3 lore raw
Quoted 10 lines · Ben Greear
Quoted 9 lines · Michal Kazior
Quoted 7 lines · Ben Greear

Possibly I've made too many local changes and gotten confused, but this code appears like it should be doing an |= instead of &= in ath10k_add_interface?

err_vdev_delete:

        ath10k_wmi_vdev_delete(ar, arvif->vdev_id);
        ar->free_vdev_map &= ~(1LL << arvif->vdev_id);
        list_del(&arvif->list);
Quoted 8 lines
Quoted 1 line

Correct. Will you submit a patch?

This is tangled in another patch that enables 64-bit vdev-maps, but I will rebase and post it shortly.

Another question in the related code: For monitor devices, we are using bit - 1 for the monitor device ID, but we do not subtract one when creating normal stations. Is this on purpose?

Ahh, nevermind that bit - 1 question, that is a local bug I introduced I think...

Thanks,
Ben

Quoted 9 lines

And, instead of checking for bit == 0, why not check for zero map before we even try the ffs call? That should work for __ffs64 as well, which is what I need for lots of vdevs...

Thanks,
Ben

Signature · 8 lines
Ben Greear <[email protected]>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k