[BUG] OTA Provider / BDX state is erroneous #23573
Comments
|
This may be due to the Also, an observation from the logs, I'm assuming that you're using an instrumented version of the code on the requestor to actually fall into this state as otherwise it wouldn't fall into this arm. But yeah in general, it can be any other failed BDX too that should have the same symptoms given the improper state the new transfer gets initialized with. |
|
Sending it to @harsha-rajendran to take an initial look |
nivi-apple
added a commit
to nivi-apple/connectedhomeip
that referenced
this issue
Feb 3, 2023
…polling again - This covers the use case where we have an image available but the requestor rejects the transfer for some reason. On the next query image the requestor is able to continue with the BDX transfer. The provider state was not being reset for this scenario due to the polling flag not being reset. - Fixes project-chip#23573
bzbarsky-apple
added a commit
that referenced
this issue
Feb 3, 2023
#24777) * Add a timer to track whether we have received BDX init after a query image was successful - Currently if an ota requester has a successful query image and an image it available but if for any reason, the ota requester doesn't send BDX init, the provider will be stuck until we reboot the resident. In order to have a fail safe, we are adding a timer that starts after query image returns image available and waits for a BDX init to come. In case BDX init doesn't come, it times out and resets state - Add code to reset state if any API fails on the provider once we prepare for BDX transfer - Stop polling when BDX transfer reset is called - Return QueryImageResponse status busy instead of general failure if the sdk is busy and gets a sexond query image so accessory can handle the error correctly and retry until the sdk is done - When we send a message, handle the case where if an error happens in sending the message close the exchange context and reset state. If the message is a status report and it succeeds, null out the exchange and reset state * Update src/darwin/Framework/CHIP/MTROTAProviderDelegateBridge.mm Co-authored-by: Boris Zbarsky <bzbarsky@apple.com> * Update src/darwin/Framework/CHIP/MTROTAProviderDelegateBridge.mm Co-authored-by: Boris Zbarsky <bzbarsky@apple.com> * Reset mStopPolling to false when we call PrepareForTransfer to start polling again - This covers the use case where we have an image available but the requestor rejects the transfer for some reason. On the next query image the requestor is able to continue with the BDX transfer. The provider state was not being reset for this scenario due to the polling flag not being reset. - Fixes #23573 * Addressed review comments * Fix the handling of the use case when hasUpdate is false we need to send the response as is - Log the error if the BDX timer fails to start --------- Co-authored-by: Boris Zbarsky <bzbarsky@apple.com>
lecndav
pushed a commit
to lecndav/connectedhomeip
that referenced
this issue
Mar 22, 2023
project-chip#24777) * Add a timer to track whether we have received BDX init after a query image was successful - Currently if an ota requester has a successful query image and an image it available but if for any reason, the ota requester doesn't send BDX init, the provider will be stuck until we reboot the resident. In order to have a fail safe, we are adding a timer that starts after query image returns image available and waits for a BDX init to come. In case BDX init doesn't come, it times out and resets state - Add code to reset state if any API fails on the provider once we prepare for BDX transfer - Stop polling when BDX transfer reset is called - Return QueryImageResponse status busy instead of general failure if the sdk is busy and gets a sexond query image so accessory can handle the error correctly and retry until the sdk is done - When we send a message, handle the case where if an error happens in sending the message close the exchange context and reset state. If the message is a status report and it succeeds, null out the exchange and reset state * Update src/darwin/Framework/CHIP/MTROTAProviderDelegateBridge.mm Co-authored-by: Boris Zbarsky <bzbarsky@apple.com> * Update src/darwin/Framework/CHIP/MTROTAProviderDelegateBridge.mm Co-authored-by: Boris Zbarsky <bzbarsky@apple.com> * Reset mStopPolling to false when we call PrepareForTransfer to start polling again - This covers the use case where we have an image available but the requestor rejects the transfer for some reason. On the next query image the requestor is able to continue with the BDX transfer. The provider state was not being reset for this scenario due to the polling flag not being reset. - Fixes project-chip#23573 * Addressed review comments * Fix the handling of the use case when hasUpdate is false we need to send the response as is - Log the error if the BDX timer fails to start --------- Co-authored-by: Boris Zbarsky <bzbarsky@apple.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Reproduction steps
Bug prevalence
Always after an unsuccessful OTA upgrade.
GitHub hash of the SDK that was being used
0b17bce
Platform
nrf, raspi, core
Platform Version(s)
No response
Anything else?
Please ignore the actual value of the software version you see in the logs.
OTA Requestor log
OTA Provider log
The text was updated successfully, but these errors were encountered: