Hi Jan,
Thank you for the answer. Did you succeed making GATT/SR/GAC/BI-01-C test as well?
I found out - thanks to Szymon Janc (sjanc) who gave an answer on the #bluez channel - that because of the patch "Add module option to enable ECRED mode" I have to enable 'enable_ecred' for the module 'bluetooth' in order to get L2CAP Enhanced Credit Based Mode support active.
Best regards,
Konstantin
Von: Ryll, Jan (GED-SDD2) <[email protected]>
Gesendet: Dienstag, 21. April 2020 14:42
An: CEE Forostyan, Konstantin; [email protected]
Betreff: RE: How to to get L2CAP Enhanced Credit Based Mode working?
?
Hi Konstantin,
we have qualified the bluez-5.52 regarding Low Energy using Services and Characteristics with normal read, write and notify but also for long characteristics.
Only problem was a bug within bluez where wrong error code was returned in case of read/write at specific offset.
The response should be "InvalidOffset" but is "InvalidValueLength". We have fixed this here and submitted the bug. But no answer if patch is applied or not.
Best regards
Jan
-----Original Message-----
From: [email protected] <[email protected]> On Behalf Of CEE Forostyan, Konstantin
Sent: Tuesday, April 21, 2020 12:45 PM
To: [email protected]
Subject: How to to get L2CAP Enhanced Credit Based Mode working?
Hi All,
I'm trying to enable L2CAP Enhanced Credit Based Mode in order to pass BT qualification test GATT/SR/GAC/BI-01-C, but it seems that something is still missing. When my BT stack receives "LE L2CAP: Enhanced Credit Connection Request (0x17)" it answers with "LE L2CAP: Command Reject (0x01) / Reason: Command not understood (0x0000)"
I'm using BlueZ 5.54 and kernel '4.20.0-yocto-standard' with the following patches taken from 'bluetooth-next':
From 145720963b6c68d0c4054112c09050995259b8f8 Mon Sep 17 00:00:00 2001
From: Luiz Augusto von Dentz <[email protected]>
Date: Mon, 2 Mar 2020 16:56:19 -0800
Subject: Bluetooth: L2CAP: Add definitions for Enhanced Credit Based Mode
From 15f02b91056253e8cdc592888f431da0731337b8 Mon Sep 17 00:00:00 2001
From: Luiz Augusto von Dentz <[email protected]>
Date: Mon, 2 Mar 2020 16:56:20 -0800
Subject: Bluetooth: L2CAP: Add initial code for Enhanced Credit Based Mode
From 4be5ca67d59d707a4b1c8608ca230ad65aa4f232 Mon Sep 17 00:00:00 2001
From: Luiz Augusto von Dentz <[email protected]>
Date: Mon, 2 Mar 2020 16:56:21 -0800
Subject: Bluetooth: L2CAP: Add module option to enable ECRED mode
From da49b602f7f75ccc91386e1274b3ef71676cd092 Mon Sep 17 00:00:00 2001
From: Luiz Augusto von Dentz <[email protected]>
Date: Wed, 25 Mar 2020 12:37:53 -0700
Subject: Bluetooth: L2CAP: Use DEFER_SETUP to group ECRED connections
Can anyone help me to enable "Enhanced Credit Based Mode", please?
Thank you and best regards,
Konstantin
Hi All,
During Bluetooth qualification tests I found out that Bluetooth stack responds with an error to a correct L2CAP_CREDIT_BASED_CONNECTION_REQ. Please review the 2 bugfixes I made in kernel in order to correct this problem.
Best regards
Konstantin
--- a/net/bluetooth/l2cap_core.c 2020-04-28 18:19:10.481887583 +0200
+++ b/net/bluetooth/l2cap_core.c 2020-04-28 18:21:41.000000000 +0200
@@ -5817,7 +5817,7 @@
if (!enable_ecred)
return -EINVAL;
- if (cmd_len < sizeof(*req) || cmd_len - sizeof(*req) % sizeof(u16)) {
+ if (cmd_len < sizeof(*req) || (cmd_len - sizeof(*req)) % sizeof(u16)) {
result = L2CAP_CR_LE_INVALID_PARAMS;
goto response;
}
@@ -5855,7 +5855,7 @@
}
result = L2CAP_CR_LE_SUCCESS;
- cmd_len -= sizeof(req);
+ cmd_len -= sizeof(*req);
num_scid = cmd_len / sizeof(u16);
for (i = 0; i < num_scid; i++) {
Hi Konstantin,
On Wed, Apr 29, 2020 at 9:52 AM CEE Forostyan, Konstantin
<[email protected]> wrote:
>
> Hi All,
>
> During Bluetooth qualification tests I found out that Bluetooth stack responds with an error to a correct L2CAP_CREDIT_BASED_CONNECTION_REQ. Please review the 2 bugfixes I made in kernel in order to correct this problem.
>
> Best regards
> Konstantin
>
>
> --- a/net/bluetooth/l2cap_core.c 2020-04-28 18:19:10.481887583 +0200
> +++ b/net/bluetooth/l2cap_core.c 2020-04-28 18:21:41.000000000 +0200
> @@ -5817,7 +5817,7 @@
> if (!enable_ecred)
> return -EINVAL;
>
> - if (cmd_len < sizeof(*req) || cmd_len - sizeof(*req) % sizeof(u16)) {
> + if (cmd_len < sizeof(*req) || (cmd_len - sizeof(*req)) % sizeof(u16)) {
> result = L2CAP_CR_LE_INVALID_PARAMS;
> goto response;
> }
> @@ -5855,7 +5855,7 @@
> }
>
> result = L2CAP_CR_LE_SUCCESS;
> - cmd_len -= sizeof(req);
> + cmd_len -= sizeof(*req);
> num_scid = cmd_len / sizeof(u16);
>
> for (i = 0; i < num_scid; i++) {
Weird that I didn't run into this when creating the test for
l2cap-tester, they look correct though so please send them as proper
patches so we can apply to bluetooth-next, also please ensure
l2cap-tester is working.
--
Luiz Augusto von Dentz
Hi Luiz,
Sorry for delay with the answer, I had a day off last Friday.
> so please send them as proper patches
I did this in a sepatare email, hopefully it is correct. This is my first time submitting a kernel patch.
> please ensure l2cap-tester is working.
Unfortunately I didn't find a way to even integrate l2cap-tester into our embedded platform, so I couldn't make any tests with it. Is it a separate tool or something that is integrated into test-runner tool? If yes, the test-runner requires qemu that we don't have.
Best regards,
Konstantin
Von: Luiz Augusto von Dentz <[email protected]>
Gesendet: Mittwoch, 29. April 2020 19:17
An: CEE Forostyan, Konstantin
Cc: [email protected]
Betreff: Re: [PATCH] Bluetooth: L2CAP: Fix errors during L2CAP_CREDIT_BASED_CONNECTION_REQ (0x17)
?
Hi Konstantin,
On Wed, Apr 29, 2020 at 9:52 AM CEE Forostyan, Konstantin
<[email protected]> wrote:
>
> Hi All,
>
> During Bluetooth qualification tests I found out that Bluetooth stack responds with an error to a correct L2CAP_CREDIT_BASED_CONNECTION_REQ. Please review the 2 bugfixes I made in kernel in order to correct this problem.
>
> Best regards
> Konstantin
>
>
> --- a/net/bluetooth/l2cap_core.c??????? 2020-04-28 18:19:10.481887583 +0200
> +++ b/net/bluetooth/l2cap_core.c??????? 2020-04-28 18:21:41.000000000 +0200
> @@ -5817,7 +5817,7 @@
>???????? if (!enable_ecred)
>???????????????? return -EINVAL;
>
> -?????? if (cmd_len < sizeof(*req) || cmd_len - sizeof(*req) % sizeof(u16)) {
> +?????? if (cmd_len < sizeof(*req) || (cmd_len - sizeof(*req)) % sizeof(u16)) {
>???????????????? result = L2CAP_CR_LE_INVALID_PARAMS;
>???????????????? goto response;
>???????? }
> @@ -5855,7 +5855,7 @@
>???????? }
>
>???????? result = L2CAP_CR_LE_SUCCESS;
> -?????? cmd_len -= sizeof(req);
> +?????? cmd_len -= sizeof(*req);
>???????? num_scid = cmd_len / sizeof(u16);
>
>???????? for (i = 0; i < num_scid; i++) {
Weird that I didn't run into this when creating the test for
l2cap-tester, they look correct though so please send them as proper
patches so we can apply to bluetooth-next, also please ensure
l2cap-tester is working.
--
Luiz Augusto von Dentz