Commit graph

13 commits

Author SHA1 Message Date
Kenshi Kawaguchi
84856f80a9 ota_tftp_download takes an optional receive_cb that will report on the status of the TFTP transfer 2016-05-28 11:27:10 +10:00
Angus Gratton
d9202af2aa Use latest upstream rboot, always build with OTA - use prebuilt rboot if
none is compiled locally.
2016-05-28 11:27:10 +10:00
Angus Gratton
1f1881a452 rboot-ota: Always put a checksum in rboot config
Means bootloader will still work if configured to verify the checksum
2016-05-28 11:27:10 +10:00
Angus Gratton
53b2b50241 rboot: Add cryptographic digest support for OTA images & SHA256 example 2016-05-28 11:27:10 +10:00
Angus Gratton
03559de5cb Move rboot_verify_image to rboot-api
Removes rboot-integration.c, removes need for clients to include rboot-integration.h
2016-05-28 11:27:10 +10:00
Angus Gratton
6eceb5843c OTA: Move OTA-aware Cache_Read_Enable to core
Otherwise images built with OTA=1 are only OTA-suitable if they also
link rboot-ota.
2016-05-28 11:27:10 +10:00
Angus Gratton
e671927bd0 OTA: Add TFTP client mode, expand ota_basic example. 2016-05-28 11:27:10 +10:00
Johan Kanflo
9c49134d61 Updated rboot to #75ca33b including the flash write bug. 2016-04-07 20:29:28 +02:00
Angus Gratton
81df978fa9 Add missing MIT license for rboot-ota 2016-03-03 08:15:21 +11:00
Angus Gratton
d5e50edc0b ota_basic example, ota-tftp support: Add links to the new OTA wiki page 2016-02-08 16:58:31 +11:00
Angus Gratton
7e73693898 Rework component makefiles to have a default <component>_ROOT
This fixes a long-standing bug where build directories sometimes
expanded many levels deeper than they should have.

Component names can now no longer contains slashes (ie the component
name for 'extras/i2c' is now just 'i2c')
2015-09-08 09:59:59 +10:00
Angus Gratton
19b8383069 Working TFTP server based OTA updates
Tested with 16MBit flash configuration, two rboot update slots.

Closes #10
2015-08-05 14:10:14 +10:00
Angus Gratton
147257efa4 Almost functional OTA support
ota_basic example can receive new image via TCP.

However - writing to flash with interrupts disabled causes data loss,
and the TCP flow is very slow to recover. Linux sender quickly ramps up
RTT timer to very long retry intervals, crippling performance &
throughput.

Running the update without the flash writes causes the data to be
received quickly, so this is definitely an issue with the time taken for
the erase cycle.

Progress towards #10
2015-07-29 16:50:23 +10:00