diff options
Diffstat (limited to 'release-notes/Colibri_T20_LinuxReleaseNotesV2.3.txt')
| -rw-r--r-- | release-notes/Colibri_T20_LinuxReleaseNotesV2.3.txt | 154 |
1 files changed, 99 insertions, 55 deletions
diff --git a/release-notes/Colibri_T20_LinuxReleaseNotesV2.3.txt b/release-notes/Colibri_T20_LinuxReleaseNotesV2.3.txt index b91e5f2..1c20fdb 100644 --- a/release-notes/Colibri_T20_LinuxReleaseNotesV2.3.txt +++ b/release-notes/Colibri_T20_LinuxReleaseNotesV2.3.txt @@ -1,11 +1,11 @@ ************************************************ -Release Notes: Linux Image V2.x for Colibri T20 +Release Notes: Linux Image V2.3 for Colibri T20 ************************************************ Contents: - U-Boot 2011.06 boot loader - Linux 3.1.10 kernel, based on NVIDIA's Linux for Tegra (L4T) release R16.5 -- Angstrom/OpenEmbedded/Yocto 2014.06/daisy/1.6 based root file system, using +- Angstrom 2014.06/OpenEmbedded daisy/Yocto 1.6 based root file system, using LXDE desktop environment ==> Flashing instructions can be found at the very bottom of this release notes @@ -55,7 +55,7 @@ Linux Image V2.3 Beta 5 (December 19, 2014) - no changes - Linux kernel: - optional ADV7180 analogue video decoder as found on ACA (formerly ACM) now - works at 25 fps + works at 25 fps - fix HDMI display controller disable state - fix pixel clock polarity - Linux user-space: @@ -122,7 +122,8 @@ Linux Image V2.3 Beta 1 (August 4, 2014) - Add SPI controller platform data for hardware chip select - Fix AC97 audio capture issue - Enable CONFIG_FHANDLE required for later systemd - - Remove synchronize_rcu() from cgroup_attach_{task|proc}() to speed-up systemd boot-up + - Remove synchronize_rcu() from cgroup_attach_{task|proc}() to speed-up + systemd boot-up - Enable VLAN functionality - Fix memset-related crashes caused by recent GCC (4.7.2) optimisations - Linux user-space @@ -218,7 +219,7 @@ Known Issues: Linux Image V2.1 Beta 1 (October 22, 2013) - U-Boot: - Fix possible uninitialised array member issue in serial number to - hexadecimal coded decimal conversion routine (e.g. as used for + hexadecimal coded decimal conversion routine (e.g. as used for /proc/cpuinfo in Linux). - SD card boot offset detection - rename shell prompt to Colibri T20 @@ -226,10 +227,10 @@ Linux Image V2.1 Beta 1 (October 22, 2013) - fix overflow in HCD conversion - Linux kernel: - Support default watchdog API (see - http://developer.toradex.com/knowledge-base/watchdog-(linux)). + http://developer.toradex.com/knowledge-base/watchdog-(linux)). - Disable magic SysRq key configuration as this is not only a potential - security thread but can cause serious trouble if for some reason UART pins - are left floating and therefore cause unexpected break conditions. + security thread but can cause serious trouble if for some reason UART pins + are left floating and therefore cause unexpected break conditions. - HDMI: Retry EDID reads up to 4 times until failing. - provision for the Analogue Camera Module - V4L2 kernel drivers configured as kernel modules @@ -270,7 +271,7 @@ Linux Image V2.1 Beta 1 (October 22, 2013) Known Issues: - LVDS aka TFTLCD resolution hard-coded in kernel to VESA VGA resolution, - comments with timings for other resolutions provided + comments with timings for other resolutions provided - LP1 suspend might fail due to pending USB interrupts - LP1 suspend might hang due to display controller issue @@ -290,7 +291,8 @@ Linux Image V2.0 (March 07, 2013) - sources git.toradex.com tag T20_LinuxImageV2.0_20130305 Known Issues: -- LVDS aka TFTLCD resolution hard-coded in kernel to VESA VGA resolution, comments with timings for other resolutions provided +- LVDS aka TFTLCD resolution hard-coded in kernel to VESA VGA resolution, + comments with timings for other resolutions provided - some windows have garbled titlebar, e.g. lxterminal - lxappearance is segfaulting upon start - LP1 suspend might fail due to pending USB interrupts @@ -316,7 +318,8 @@ Linux Image V2.0 Beta 2 (January 29, 2013) - sources git.toradex.com tag T20_LinuxImageV2.0Beta2_20130129 Known Issues: -- LVDS aka TFTLCD resolution hard-coded in kernel to VESA VGA resolution, comments with timings for other resolutions provided +- LVDS aka TFTLCD resolution hard-coded in kernel to VESA VGA resolution, + comments with timings for other resolutions provided - some windows have garbled titlebar, e.g. lxterminal - lxappearance is segfaulting upon start - LP1 suspend might fail due to pending USB interrupts @@ -330,38 +333,53 @@ Linux Image V2.0 Beta 1 (December 18, 2012) - fix NVIDIA partition table parsing - fix Ethernet detection fault - Linux kernel: - - added one wire platform data (conditional, CONFIG_W1_MASTER_TEGRA must be enabled) - - implement analogue input AD0 aka ANALOG_IN0 as voltage_now and AD1 aka ANALOG_IN1 as temp value (see /sys/class/power_supply/colibri_t20-analog_inputs) + - added one wire platform data (conditional, CONFIG_W1_MASTER_TEGRA must be + enabled) + - implement analogue input AD0 aka ANALOG_IN0 as voltage_now and AD1 aka + ANALOG_IN1 as temp value (see + /sys/class/power_supply/colibri_t20-analog_inputs) - unified Ethernet initialisation - GPIO keys implementation - USB OTG support - USB gadget using Android function stack - USB full speed limiting (see usb_high_speed kernel command line argument) - USB device now reports 2 mA VBUS draw - - revise Ethernet MAC address assignment: should now use MAC from EEPROM if set and further handles up to two instances of custom user MACs (2nd one with a 0x100000 offset), this way customer does not have to put EEPROM on a secondary Ethernet on carrier board and still gets a valid official MAC address from us (e.g. analogous to how we do it on our Protea carrier board) + - revise Ethernet MAC address assignment: should now use MAC from EEPROM if + set and further handles up to two instances of custom user MACs (2nd one + with a 0x100000 offset), this way customer does not have to put EEPROM on a + secondary Ethernet on carrier board and still gets a valid official MAC + address from us (e.g. analogous to how we do it on our Protea carrier board) - fix VESA VGA graphics aka too much TuxTM issue - - using full PMIC power off which decreases power consumption but reset won't force a re-boot anymore + - using full PMIC power off which decreases power consumption but reset won't + force a re-boot anymore - implement Video-DAC power save for Iris - LP1 suspend support (Colibri T20 512 MB on Iris draws around 180 mW) - re-worked function tri-stating - irq 68 nobody cared workaround - oe-core: - L4T release 16-r2 userspace binaries - - update procedure: use loader.nb0 as the target binary, add .bat file for flashing from Windows, improved parameter checks, + - update procedure: use loader.nb0 as the target binary, add .bat file for + flashing from Windows, improved parameter checks - configure RNDIS with Androids USB gadget - - dropbear: disable reverse DNS lookups, prevents timeout if no DNS Server is available + - dropbear: disable reverse DNS lookups, prevents timeout if no DNS Server is + available - add gstreamer plugins, nvvidconv Nvidia colorspace conversion, mp3 helpers - - xinput-calibrator: make the menu entry behaviour equal to the one time calibration on first boot - - added firmware for Ralink WiFi chipsets, e.g. WL250N USB WiFi stick from Ambicom + - xinput-calibrator: make the menu entry behaviour equal to the one time + calibration on first boot + - added firmware for Ralink WiFi chipsets, e.g. WL250N USB WiFi stick from + Ambicom - add the rootfs builddate to /etc/issue -> Welcome banner - - add license information generated from oe-core and merge this with the Nvidia licenses in the common folder /usr/share/common-licenses - - kernel recipe: remove all fiddling with the defconfig, compile the kernel as it is in git + - add license information generated from oe-core and merge this with the + Nvidia licenses in the common folder /usr/share/common-licenses + - kernel recipe: remove all fiddling with the defconfig, compile the kernel + as it is in git - header files for OpenGL, EGL, OpenMAX: taken from www.khronos.org - firefox: set develper.toradex.com as the homepage - sources git.toradex.com tag T20_LinuxImageV2.0Beta1_20121218 Known Issues: -- LVDS aka TFTLCD resolution hard-coded in kernel to VESA VGA resolution, comments with timings for other resolutions provided +- LVDS aka TFTLCD resolution hard-coded in kernel to VESA VGA resolution, + comments with timings for other resolutions provided - some windows have garbled titlebar, e.g. lxterminal - lxappearance is segfaulting upon start - LP1 suspend might fail due to pending USB interrupts @@ -369,14 +387,18 @@ Known Issues: ******************************************************************************** Linux Image V2.0 Alpha 2 (October 19, 2012) -- update to L4T R16, !!!THIS REQUIRED TO CHANGE THE CALLING CONVENTION TO HARD FLOAT!!! +- update to L4T R16, !!!THIS REQUIRED TO CHANGE THE CALLING CONVENTION TO HARD + FLOAT!!! - U-Boot: - - use a generic approach to get the MTD partitions to ease changes to the partition layout - (Initial partition table parsing courtesy of Mitja \8Apes from LXNAV) + - use a generic approach to get the MTD partitions to ease changes to the + partition layout (Initial partition table parsing courtesy of Mitja \8Apes + from LXNAV) - increase kernel partition size to 8 MB - move the init output from the framebuffer console to the serial output - Linux kernel: - - implement analogue input AD0 aka ANALOG_IN0 as voltage_now and AD1 aka ANALOG_IN1 as temp value. See /sys/class/power_supply/colibri_t20-analog_inputs. + - implement analogue input AD0 aka ANALOG_IN0 as voltage_now and AD1 aka + ANALOG_IN1 as temp value. See + /sys/class/power_supply/colibri_t20-analog_inputs. - PWM B, PWM C, PWM D now accessible from userspace, /sys/class/leds/ - IPv6 enabled - USB audio enabled @@ -384,11 +406,13 @@ Linux Image V2.0 Alpha 2 (October 19, 2012) - oe-core: - updated the different meta-layers - change machine configuration to use hard float - - change OpenGL ES nvsamples to use hard float and not using pre-compiled shaders + - change OpenGL ES nvsamples to use hard float and not using pre-compiled + shaders - sources git.toradex.com tag T20_LinuxImageV2.0Alpha2_20121019 Known Issues: -- LVDS aka TFTLCD resolution hard-coded in kernel to VESA VGA resolution, comments with timings for other resolutions provided +- LVDS aka TFTLCD resolution hard-coded in kernel to VESA VGA resolution, + comments with timings for other resolutions provided - some gstreamer plugins, notably mp3decoding, h264encoding not working - some windows have garbled titlebar, e.g. lxterminal - lxappearance is segfaulting upon start @@ -396,42 +420,51 @@ Known Issues: - SD card boot copies only 4 MB from kernel (see further below for workaround) Calling Convention hard float: -The way floats are passed in subroutine calls can be either in processor registers or -in the floating point coprocessor registers. -The processor register approach is used when gcc is called with -mfloat-abi=softfp or -mfloat-abi=soft +The way floats are passed in subroutine calls can be either in processor +registers or in the floating point coprocessor registers. +The processor register approach is used when gcc is called with +-mfloat-abi=softfp or -mfloat-abi=soft The FPU register approach is used when gcc is called with -mfloat-abi=hard -These two methods are not compatible with each other, all programs and libraries must be compiled -either with softfp/soft or with hard. -Thus all code used with Alpha2 must be recompiled. If you start a program which uses soft calling -convention you will get a 'No such file or directory' error, despite that the file exists and has executable permissions. +These two methods are not compatible with each other, all programs and libraries +must be compiled either with softfp/soft or with hard. +Thus all code used with Alpha2 must be recompiled. If you start a program which +uses soft calling convention you will get a 'No such file or directory' error, +despite that the file exists and has executable permissions. ******************************************************************************** Linux Image V2.0 Alpha 1 (August 9, 2012) - migrated from Vibrante 1.1 to L4T R15 - migrated from OpenEmbedded classic to OpenEmbedded core - power management (frequency and voltage scaling) enabled -- touch calibration applied after boot, missing configuration triggers calibrator launch -- USB high speed activated by default, change if your baseboard only supports low and full speed +- touch calibration applied after boot, missing configuration triggers + calibrator launch +- USB high speed activated by default, change if your baseboard only supports + low and full speed - sources git.toradex.com tag T20_LinuxImageV2.0Alpha1_20120808 Known Issues: -- sometimes postinsts script not executed at first boot resulting in no icons and no fonts on graphical screen - workaround: manually execute /etc/init.d/run-postinsts +- sometimes postinsts script not executed at first boot resulting in no icons + and no fonts on graphical screen, workaround: manually execute + /etc/init.d/run-postinsts - LVDS aka TFTLCD resolution hardcoded in kernel to VESA VGA resolution - some gstreamer plugins, notably mp3 not working -- X server needs ugly hack not to segfault, some regions are not redrawn nicely when hoovered over by mouse +- X server needs ugly hack not to segfault, some regions are not redrawn nicely + when hoovered over by mouse - some windows have garbled titlebar, e.g. lxterminal - Firefox not yet working and thus not included ******************************************************************************** Notes: -- 115200 baud serial debug console on UARTA aka FFUART (lower X25 on EvalBoards resp. X27 on V3.1a depending on JP17/19, X13 on Iris) without any handshake +- 115200 baud serial debug console on UARTA aka FFUART (lower X25 on EvalBoards + resp. X27 on V3.1a depending on JP17/19, X13 on Iris) without any handshake - user account: root with empty password - YAFFS2 flash root file system - opkg package manager - HDMI aka DVI-D: not all pixel clock frequencies allow for HDMI audio -- CRT: the T20 internal video DAC output is unsupported by NVIDIA's L4T Linux SW stack -- Linux host used for recovery mode flashing requires proper 32-bit execution environment (e.g. install ia32-libs on 64-bit Ubuntu) +- CRT: the T20 internal video DAC output is unsupported by NVIDIA's L4T Linux SW + stack +- Linux host used for recovery mode flashing requires proper 32-bit execution + environment (e.g. install ia32-libs on 64-bit Ubuntu) Working on latest image: - U-Boot boot loader @@ -494,19 +527,23 @@ NAND Boot: Colibri T20 256MB V1.1b (Sample), Colibri T20 512MB V1.1c (Sample), C ***************************************************************************************************************************************************************************************** Initial Preparation: -- download and extract Colibri_T20_LinuxImageVx.yz.tar.bz2 package with root permissions +- download and extract Colibri_T20_LinuxImageVx.yz.tar.bz2 package with root + permissions Flash NAND: - EvalBoard V3.1a - - connect USB B (X29) or USB micro A/B (X30) to development workstation using regular or micro USB cable + - connect USB B (X29) or USB micro A/B (X30) to development workstation using + regular or micro USB cable - the jumper JP2 must be open - press and hold recovery mode switch (SW9) during power up - Iris V1.1a - - connect USB micro A/B (X12) to development workstation using regular micro USB cable + - connect USB micro A/B (X12) to development workstation using regular micro + USB cable - short pin 1-2 of JP1 on Iris for RECOVERY (other carrier boards see http://developer.toradex.com/knowledge-base/txx-recovery-mode) - make sure that the carrier board supports USB high speed - - Iris & EvalBoard V3.1a OK, for other carrier boards see http://developer.toradex.com/knowledge-base/usb-2.0-high-speed-%28480mbps%29 ) + - Iris & EvalBoard V3.1a OK, for other carrier boards see +http://developer.toradex.com/knowledge-base/usb-2.0-high-speed-%28480mbps%29 ) - apply power or reset - run update.sh script - use -r 512 option for 512 MB RAM modules @@ -518,7 +555,8 @@ Boot from NAND: - Iris V1.1a - do no longer short any pins of JP1 - power cycle -- upon first boot post-install scripts are run which takes around 2 minutes, please be patient +- upon first boot post-install scripts are run which takes around 2 minutes, + please be patient ******************************************************************************************************************************************************************************************************************************************************************* @@ -527,21 +565,26 @@ SD Card Boot: Colibri T20 256MB V1.1b (Sample), Colibri T20 512MB V1.1c (Sample) Initial Preparation: - EvalBoard V3.1a - - ground the following user extension (X3) pins to select the SD boot bootstrap option: A6, B6, C5 and C6 + - ground the following user extension (X3) pins to select the SD boot + bootstrap option: A6, B6, C5 and C6 - insert SD card into SD card/MMC SDIO-socket (X15) - Iris V1.1a - - populate 100 ohm resistors R109, R111, R112 and R113 close to micro SD card holder (X10, see bottom assembly drawing fig.7 page 21 of Iris data sheet) + - populate 100 ohm resistors R109, R111, R112 and R113 close to micro SD card + holder (X10, see bottom assembly drawing fig.7 page 21 of Iris data sheet) - populate boot mode header (JP1) - insert micro SD card into micro SD card holder (X10) -- download and extract Colibri_T20_LinuxImageVx.yz.tar.bz2 package with root permissions +- download and extract Colibri_T20_LinuxImageVx.yz.tar.bz2 package with root + permissions Flash SD Card: - EvalBoard V3.1a - - connect USB B (X29) or USB micro A/B (X30) to development workstation using regular or micro USB cable + - connect USB B (X29) or USB micro A/B (X30) to development workstation using + regular or micro USB cable - the jumper JP2 must be open - press and hold recovery mode switch (SW9) during power up - Iris V1.1a - - connect USB micro A/B (X12) to development workstation using regular micro USB cable + - connect USB micro A/B (X12) to development workstation using regular micro + USB cable - short all 3 pins of JP1 for RECOVERY and SD_BOOT - apply power or reset - run update.sh script with -b hsmmc option @@ -554,6 +597,7 @@ Boot from SD Card: - Iris V1.1a - short pin 2-3 of JP1 for SD_BOOT - power cycle -- upon first boot post-install scripts are run which takes around 2 minutes, please be patient +- upon first boot post-install scripts are run which takes around 2 minutes, + please be patient -Toradex, March 25, 2015 +Toradex, May 29, 2015 |