From 3b51fc2c5e06e35a8058bf1a6ac67b38c5f2e25f Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ricardo=20Ca=C3=B1uelo?= <ricardo.canuelo@collabora.com> Date: Thu, 17 Dec 2020 10:55:07 +0100 Subject: [PATCH] boards/chromebooks: Update board status for coral, grunt, rammus MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Ricardo Cañuelo <ricardo.canuelo@collabora.com> --- .../chromebooks/boards/common_issues.md | 60 +++++++++++++++++++ content/boards/chromebooks/boards/coral.md | 19 ++++-- content/boards/chromebooks/boards/grunt.md | 28 ++++++--- content/boards/chromebooks/boards/rammus.md | 4 +- 4 files changed, 94 insertions(+), 17 deletions(-) create mode 100644 content/boards/chromebooks/boards/common_issues.md diff --git a/content/boards/chromebooks/boards/common_issues.md b/content/boards/chromebooks/boards/common_issues.md new file mode 100644 index 0000000..52b529d --- /dev/null +++ b/content/boards/chromebooks/boards/common_issues.md @@ -0,0 +1,60 @@ +--- +title: Common issues +--- + +This is a list of the issues we found that were common for all the +Chromebooks tested in the lab. + + +## Issues related to Servo v4 + +The Servo v4 debugging interface causes different issues depending on +the Chromebook board and the lab setup. + +### Network interface issues + +#### R8152 driver not reliable at Gigabit speeds + +The Realtek R8152 USB-Eth adapter can be found in the Servo v4 and in +other standalone dongles. + +The R8152 Ethernet driver in Depthcharge is not reliable when working at +Gigabit speeds. It's recommeded to configure the link as Fast Ethernet +(100 Mbps) when booting over TFTP. + + +#### Ethernet frame reception issues + +Ethernet frame reception may fail and hang Depthcharge when using the +Servo v4 as an USB-Eth adapter (error: `R8152: Bulk read error`). Bigger +file transfers make this more likely to happen, although it's not +completely deterministic. + + +### Stability issues + +#### Debugging interface unstable when using Servo v4 connected to power supply + +Observed in: + +- `coral`: R64-10068.111.0.serial +- `rammus`: R72-11275.110.0.serial +- `octopus`: R72-1197.144.0.serial + +with Servo v4 FW 2.3.22 and 2.4.35. + +When the Servo v4 is connected to a power supply the debugging +interface becomes unstable. Some of the effects seen: + + - Charging is unreliable. + - Serial consoles stop working. The following message appears in the + kernel log: + +``` +google ttyUSB4: usb_serial_generic_write_bulk_callback - urb stopped: -32 +``` + - Ethernet data transfer through the Servo fails. + - Cr50 commands lost sometimes. + +Basically, all communication going through the Servo starts failing +intermittently. diff --git a/content/boards/chromebooks/boards/coral.md b/content/boards/chromebooks/boards/coral.md index 8e17c7a..a2684f8 100644 --- a/content/boards/chromebooks/boards/coral.md +++ b/content/boards/chromebooks/boards/coral.md @@ -18,13 +18,17 @@ and devices, but they are based on Intel Apollo Lake 64 bit CPUs The Collabora LAVA lab contains the following `coral` devices: - ASUS Chromebook C523NA - - `asus-C523NA-A20057-coral` + - [`asus-C523NA-A20057-coral-cbg-0`](https://lava.collabora.co.uk/scheduler/device/asus-C523NA-A20057-coral-cbg-0) + - [`asus-C523NA-A20057-coral-cbg-1`](https://lava.collabora.co.uk/scheduler/device/asus-C523NA-A20057-coral-cbg-1) + - [`asus-C523NA-A20057-coral-cbg-2`](https://lava.collabora.co.uk/scheduler/device/asus-C523NA-A20057-coral-cbg-2) + - [`asus-C523NA-A20057-coral-cbg-3`](https://lava.collabora.co.uk/scheduler/device/asus-C523NA-A20057-coral-cbg-3) + + ### Debugging interfaces `coral` boards have been flashed and tested with both [SuzyQ and Servo -v4](../../01-debugging_interfaces) interfaces. No issues reported so -far. +v4](../../01-debugging_interfaces) interfaces. In an Asus C523N, the debug port is the USB-C port in the left side (also used for the power supply). @@ -36,9 +40,12 @@ supported by Depthcharge (R8152). #### Known issues -The R8152 Ethernet driver in Depthcharge doesn't seem reliable when -working at Gigabit speeds. It's recommeded to configure the link as Fast -Ethernet when booting over TFTP. + - [`coral-1`](https://lava.collabora.co.uk/scheduler/device/asus-C523NA-A20057-coral-cbg-1) + fails to boot in around 5% of the attempts. One of the Coreboot + stages resets the board and then it never comes out of reset (last + console message: `CSE timed out. Resetting`). + +See also [Common issues](../common_issues.md). ### Example kernel command line arguments diff --git a/content/boards/chromebooks/boards/grunt.md b/content/boards/chromebooks/boards/grunt.md index 152ea8d..0c71b63 100644 --- a/content/boards/chromebooks/boards/grunt.md +++ b/content/boards/chromebooks/boards/grunt.md @@ -41,14 +41,26 @@ supported by Depthcharge (R8152). #### Known issues -The R8152 Ethernet driver in Depthcharge doesn't seem reliable when -working at Gigabit speeds. It's recommeded to configure the link as Fast -Ethernet when booting over TFTP. - -Depending on the firmware version of the Servo v4 interface used, some -issues may be present, such as the loss of the cr50 serial console when -issuing a cold reset or Depthcharge hanging when receiving network -packets. + - When using a Servo v4 as a debugging interface, the Cr50 interface + shuts down after issuing a `cold_reset:on` command. However, this + doesn't seem to happen if the Servo v4 is connected to a power + supply and charging the Chromebook. + + - In some setups the serial console gets stuck and freezes during one + of the Coreboot stages. Closing and reopening the serial terminal + recovers the problem, since the Chromebook keeps running. Most of + the times this happens is at this point during the boot process: + + ``` + POST: 0x40 + agesawrapper_amdinitpost() entry + DRAM clear on reset: Keep + variant_mainboard_read_spd SPD index 9 + CBFS: 'Master Header Locator' located CBFS at [df0000:ffffc0) + CBFS: Locating 'spd.bin' + CBFS: Found @ offset 79bc0 size 2000 + ``` +See also [Common issues](../common_issues.md). ### Example kernel command line arguments diff --git a/content/boards/chromebooks/boards/rammus.md b/content/boards/chromebooks/boards/rammus.md index aeb606e..05d098c 100644 --- a/content/boards/chromebooks/boards/rammus.md +++ b/content/boards/chromebooks/boards/rammus.md @@ -27,9 +27,7 @@ supported by Depthcharge (R8152). #### Known issues -The R8152 Ethernet driver in Depthcharge doesn't seem reliable when -working at Gigabit speeds. It's recommeded to configure the link as Fast -Ethernet when booting over TFTP. +See also [Common issues](../common_issues.md). ### Example kernel command line arguments -- GitLab