Mdns: separate query responders for advertisers #7615

cecille · 2021-06-14T18:40:04Z

Problem

The current setup uses a single query responder for commissionable
and operational and all the records are cleared together and added
in a large group. This makes it challenging to run commissionable
and operational at the same time since they can't be cleared
individually and because there are duplicate records between
operational and commissionable.
Fixes #7175

Change overview

This commit changes the code to use one query responder per advertisement type and gets the
ResponseSender to loop through the query responders to filter for answers.

Testing

Unit tests added for multiple query responders. Single responder tests still pass.
Manually tested on M5. avahi, chip-device-ctrl both show correct records, confirmed with wireshark
when the clear calls are removed, can have operational and commissionable running with no duplicate records.

todo · 2021-06-14T18:40:06Z

When multi-admin is enabled, commissionable does not need to be cleared here.

connectedhomeip/src/lib/mdns/Advertiser_ImplMinimalMdns.cpp

Lines 214 to 224 in 5e33f3d

    
           // TODO: When multi-admin is enabled, commissionable does not need to be cleared here. 
        
           mQueryResponderAllocatorOperational.Clear(); 
        
           mQueryResponderAllocatorCommissionable.Clear(); 
        
           mQueryResponderAllocatorCommissioner.Clear(); 
        
           char nameBuffer[64] = ""; 
        
           /// need to set server name 
        
           ReturnErrorOnFailure(MakeInstanceName(nameBuffer, sizeof(nameBuffer), params.GetPeerId())); 
        
           FullQName operationalServiceName = mQueryResponderAllocatorOperational.AllocateQName("_chip", "_tcp", "local"); 
        
           FullQName operationalServerName  = mQueryResponderAllocatorOperational.AllocateQName(nameBuffer, "_chip", "_tcp", "local");

This comment was generated by todo based on a `TODO` comment in `5e33f3d` in #7615. cc @cecille.

todo · 2021-06-14T18:40:09Z

When multi-admin is enabled, operational does not need to be cleared here.

connectedhomeip/src/lib/mdns/Advertiser_ImplMinimalMdns.cpp

Lines 282 to 289 in 5e33f3d

    
           // TODO: When multi-admin is enabled, operational does not need to be cleared here. 
        
           mQueryResponderAllocatorOperational.Clear(); 
        
           mQueryResponderAllocatorCommissionable.Clear(); 
        
           mQueryResponderAllocatorCommissionable.Clear(); 
        
           // TODO: need to detect colisions here 
        
           char nameBuffer[64] = ""; 
        
           size_t len          = snprintf(nameBuffer, sizeof(nameBuffer), ChipLogFormatX64, GetRandU32(), GetRandU32());

This comment was generated by todo based on a `TODO` comment in `5e33f3d` in #7615. cc @cecille.

cecille · 2021-06-14T18:46:37Z

@chrisdecenzo @sweetymhaiske - This PR follows the Clear() patterns used on master currently, but I think it conflicts with
#7397.

Chris - I'm not totally clear about what the problem was for TVs - can you elaborate and check this PR to see if this is going to cause you issues?

chrisdecenzo

please see comments on not clearing unrelated service types.

src/lib/mdns/Advertiser_ImplMinimalMdns.cpp

src/lib/mdns/minimal/ResponseSender.h

woody-apple · 2021-06-16T04:02:12Z

/rebase

woody-apple · 2021-06-17T01:34:28Z

/rebase

The current setup uses a single query responder for commissionable and operational and all the records are cleared together and added in a large group. This makes it challenging to run commissionable and operational at the same time since they can't be cleared individually and because there are duplicate records between operational and commissionable. This commit changes the code to use one query responder per advertisement type and gets the ResponseSender to loop through the query responders to filter for answers.

Co-authored-by: chrisdecenzo <[email protected]>

…ry_responders

woody-apple · 2021-06-21T22:22:45Z

Real test failures here

stale · 2021-06-28T22:41:38Z

This pull request has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.

woody-apple · 2021-06-29T18:18:31Z

@saurabhst @msandstedt @mspang @jmartinez-silabs @Damian-Nordic ?

Represents 5 multi-admin fabrics for operational, commissionable and commissioner.

todo · 2021-06-29T23:57:53Z

(cecille): Template this and set appropriately. Please see issue #8000.

connectedhomeip/src/lib/mdns/minimal/ResponseSender.h

Lines 92 to 99 in e74cfad

    
           // TODO(cecille): Template this and set appropriately. Please see issue #8000. 
        
           static constexpr size_t kMaxQueryResponders = 7; 
        
           ResponseSender(ServerBase * server) : mServer(server) {} 
        
           CHIP_ERROR AddQueryResponder(QueryResponderBase * queryResponder); 
        
           /// Send back the response to a particular query 
        
           CHIP_ERROR Respond(uint32_t messageId, const QueryData & query, const chip::Inet::IPPacketInfo * querySource);

This comment was generated by todo based on a `TODO` comment in `e74cfad` in #7615. cc @cecille.

todo · 2021-06-30T12:11:25Z

(cecille): Fix this test once #8000 gets resolved.

connectedhomeip/src/lib/mdns/minimal/tests/TestResponseSender.cpp

Lines 358 to 368 in b230276

    
           // TODO(cecille): Fix this test once #8000 gets resolved. 
        
           CommonTestElements common1(inSuite, "test1"); 
        
           CommonTestElements common2(inSuite, "test2"); 
        
           CommonTestElements common3(inSuite, "test3"); 
        
           CommonTestElements common4(inSuite, "test4"); 
        
           CommonTestElements common5(inSuite, "test5"); 
        
           CommonTestElements common6(inSuite, "test6"); 
        
           CommonTestElements common7(inSuite, "test7"); 
        
           CommonTestElements common8(inSuite, "test8"); 
        
           ResponseSender responseSender(&common1.server);

This comment was generated by todo based on a `TODO` comment in `b230276` in #7615. cc @cecille.

…ry_responders

* add electrical measurement cluster to controller-clusters.zap Only a selection of the optional clusters is enabled. mandatory: - measurement type optional: - total active power - rms voltage - rms voltage min - rms voltage max - rms current - rms current min - rms current max - active power - active power min - active power max * add electrical measurement cluster to zap_cluster_list.py * add electrical measurement cluster to all-clusters-app * add electrical measurement cluster to all-cluster-app/esp32 * add app helpers for electrical measurement cluster * fix all cluster app build * regen all * Regenerating * Restyled by whitespace * Restyled by clang-format * Regenerating * Restyled by whitespace * Restyled by google-java-format * [fix] Wrong callback passed to WriteAttribute in Python[TE4] (#8019) * Fix callback typr * Run codegen * Mdns: _chip* -> _matter*, remove 0's on subtypes (#7805) * Mdns: _chip* i-> _matter*, remove 0's on subtypes The second ballot changed the format of the subtypes as well as fixed the service name prefix from _chip* to _matter*. NOTE FOR DEVs: Updating past this change will require updating the devices and controllers as otherwise they will be unable to find each other. * Darwin - use local kLocalDomain var. It includes a . and the one in the ServiceNaming doesn't. * Trying again for Darwin. * Fix host name. * Fix small error on merge * Use a dedicated chip-types.xml file into src/app/zap-templates/zcl (#7873) * Update chip-types.xml * Update gen/ folders * Revert "Mark UDP socket as ready for write in send msg. (#7927)" (#8027) Turns out the code causes that subsequent select()-s never wait so another solution for the original issue must be prepared. This reverts commit 8c55523. * [ZAP] Add isAnalog instead of isDiscreteType to account for string types (#7821) * Mdns: separate query responders for advertisers (#7615) * Mdns: separate query responders for advertisers The current setup uses a single query responder for commissionable and operational and all the records are cleared together and added in a large group. This makes it challenging to run commissionable and operational at the same time since they can't be cleared individually and because there are duplicate records between operational and commissionable. This commit changes the code to use one query responder per advertisement type and gets the ResponseSender to loop through the query responders to filter for answers. * Apply suggestions from code review Co-authored-by: chrisdecenzo <[email protected]> * Restyled by clang-format * Fix tests - they were not testing right. * Restyled by clang-format * Update number of responders Represents 5 multi-admin fabrics for operational, commissionable and commissioner. * Whoops - broke a test with my last "minor" change. * Lighten the stack load for the failing test. Co-authored-by: chrisdecenzo <[email protected]> Co-authored-by: Restyled.io <[email protected]> * [nrfconnect] Stop using cooperative threads (#8025) nRF Connect platform used cooperative threads, that is threads that cannot be preempted, to workaround some synchronization issues in the past, but it's no longer needed and is risky. The latter was revealed by #7927, which caused that `select()` never waits (needs to be fixed in another PR). Signed-off-by: Damian Krolik <[email protected]> * IPV6-only test app for esp32 (examples/ipv6only-app) (#7289) Adds RPCs that conform to NDM/GDM WiFi capabilities. Disables IPV4 DHCP client. Assigns link-local IPV6 address. Starts a UDP echo server on an IPV6 UDP socket. Test: Ran on M5Stack and verified it scanned and connected to wifi, after connected verified it echoed UDP messages. * [ESP32] fix compile error for persistent-storage app (#7940) * [ESP32] fix compile error for persistent-storage app * add CI for all ESP examples * Consolidate error types (#7916) * Consolidate error types #### Problem CHIP currently has `CHIP_ERROR`, `INET_ERROR`, `BLE_ERROR`, `ASN1_ERROR`, `System::Error`. Since they're all integers underneath, there is no type checking, and existing code routinely treats them as interchangeable. There is also existing code that uses `int` and `int32_t`. Each also has a `…_NO_ERROR` code which is independently configurable, but existing code assumes they are equal, and there is exising code that assumes the C convention that the value is 0. #### Change overview Fixes #7688 Unify chip errors This change is largely a mechanical replacement of type and some error code names, with a few cleanups of weak typing. Consolidated error types to one, `CHIP_ERROR`. Removed configurability of `CHIP_NO_ERROR`; it must be 0. Added `CHIP_ERROR_FORMAT` to replace hardcoded `PRI…32` in `printf()` format strings. Added `CHIP_ERROR_SENTINEL` to replace the use of `CHIP_ERROR_MAX` as a sentinel. Renamed or consolidated the following errors, which have the same meaning. Merely similar codes are left alone. This change contains compatibility defintitions to avoid breaking PRs in flight, which will be removed in a followup. | before | after | |:---------------------------------------|:------------------------------------| | ASN1_ERROR | CHIP_ERROR | | ASN1_NO_ERROR | CHIP_NO_ERROR | | BLE_ERROR | CHIP_ERROR | | BLE_NO_ERROR | CHIP_NO_ERROR | | BLE_ERROR_BAD_ARGS | CHIP_ERROR_INVALID_ARGUMENT | | BLE_ERROR_INCORRECT_STATE | CHIP_ERROR_INCORRECT_STATE | | BLE_ERROR_MESSAGE_INCOMPLETE | CHIP_ERROR_MESSAGE_INCOMPLETE | | BLE_ERROR_NOT_IMPLEMENTED | CHIP_ERROR_NOT_IMPLEMENTED | | BLE_ERROR_NO_ENDPOINTS | CHIP_ERROR_ENDPOINT_POOL_FULL | | BLE_ERROR_NO_MEMORY | CHIP_ERROR_NO_MEMORY | | BLE_ERROR_OUTBOUND_MESSAGE_TOO_BIG | CHIP_ERROR_OUTBOUND_MESSAGE_TOO_BIG | | BLE_ERROR_RECEIVED_MESSAGE_TOO_BIG | CHIP_ERROR_INBOUND_MESSAGE_TOO_BIG | | CHIP_SYSTEM_NO_ERROR | CHIP_NO_ERROR | | CHIP_SYSTEM_ERROR_ACCESS_DENIED | CHIP_ERROR_ACCESS_DENIED | | CHIP_SYSTEM_ERROR_BAD_ARGS | CHIP_ERROR_INVALID_ARGUMENT | | CHIP_SYSTEM_ERROR_NOT_SUPPORTED | CHIP_ERROR_UNSUPPORTED_CHIP_FEATURE | | CHIP_SYSTEM_ERROR_NO_MEMORY | CHIP_ERROR_NO_MEMORY | | CHIP_SYSTEM_ERROR_REAL_TIME_NOT_SYNCED | CHIP_ERROR_REAL_TIME_NOT_SYNCED | | CHIP_SYSTEM_ERROR_UNEXPECTED_EVENT | CHIP_ERROR_UNEXPECTED_EVENT | | CHIP_SYSTEM_ERROR_UNEXPECTED_STATE | CHIP_ERROR_INCORRECT_STATE | | INET_ERROR | CHIP_ERROR | | INET_NO_ERROR | CHIP_NO_ERROR | | INET_ERROR_BAD_ARGS | CHIP_ERROR_INVALID_ARGUMENT | | INET_ERROR_INBOUND_MESSAGE_TOO_BIG | CHIP_ERROR_INBOUND_MESSAGE_TOO_BIG | | INET_ERROR_INCORRECT_STATE | CHIP_ERROR_INCORRECT_STATE | | INET_ERROR_MESSAGE_TOO_LONG | CHIP_ERROR_MESSAGE_TOO_LONG | | INET_ERROR_NO_CONNECTION_HANDLER | CHIP_ERROR_NO_CONNECTION_HANDLER | | INET_ERROR_NO_ENDPOINTS | CHIP_ERROR_ENDPOINT_POOL_FULL | | INET_ERROR_NOT_IMPLEMENTED | CHIP_ERROR_NOT_IMPLEMENTED | | INET_ERROR_NOT_SUPPORTED | CHIP_ERROR_UNSUPPORTED_CHIP_FEATURE | | INET_ERROR_NO_MEMORY | CHIP_ERROR_NO_MEMORY | | INET_ERROR_CONNECTION_ABORTED | CHIP_ERROR_CONNECTION_ABORTED | #### Testing Modified some unit tests to account for consolidated error codes. Built locally with `CHIP_ERROR` hardcoded to `int32_t`, `uint32_t`, `int16_t`, and `uint16_t` to shake out mistaken type assumptions. * Build fixes * Darwin * Doxygen * Fixes * Change default CHIP_CONFIG_ERROR_TYPE back to `int32_t`. * Fix CHIP_ERROR_INBOUND_MESSAGE_TOO_BIG message text. * C++-style casts. * fix: CHIPError.h uses stdint.h * zap * fix for Mbed merge * Throw exception instead of return 0 for get-fabricid in case of error (#7884) * Implement User Directed Commissioning - Phase 1 (#8034) * Updated CHIP Certs Implementation to Use Spans instaed of Pointer/Len Pairs. (#7599) Specifically, added the following FixedSpan types: - CertificateKeyId - P256ECDSASignatureSpan - P256PublicKeySpan Also redefined mString member in the ChipRDN structure to be ByteSpan. * Zap use const expr for clusters commands attributes (#7946) * Remove duplicated command ids and attribute ids definitions from the encoder-src.zapt template Add ids/Commands.zapt template Update template to use Commands.h and Clusters.h * Fix XML definitions typos for attributes name * Add ids/Commands.h to the gen folder of src/app/common * Update log messages for session expiry (#8017) * Update log messages for session expiry * update format specifier * Remove client-command-macro.h-aa80d825 from examples/bridge-app/bridge-common/gen (#8026) * Fix Ble endpoint pool FULL on macOS controller (#7806) * * Move naming qpg6100 to qpg (#7936) * Removal Env overrides sdk root and target IC * fix build error in TCP.cpp (#8072) Include <limits> to satisfy the proper use of std::numeric_limits. * Add lighting example for Mbed OS (#8048) * Add back structs, bitmaps and enums of the Messaging cluster into ami… (#8061) * Add back structs, bitmaps and enums of the Messaging cluster into ami.xml * Update gen/ folders * Deprecated ASN1Writer::Finalize() Method. (#8056) * Split KeyValueStoreManager.java into interface and impl, and add Set() in JNI. (#8046) * [Test] Add Test_TC_DM_3_1 (#8067) * Add Test_TC_DM_3_1 to src/app/tests/suites with all tests disabled since the attributes are not present into commissioning.xml * Update generated tests * Followup #7873 - Replace emberAfPutInt8uInResp(ZCL_DEFAULT_RESPONSE_COMMAND_ID) by emberAfPutInt32uInResp(ZCL_DEFAULT_RESPONSE_COMMAND_ID) in src/app/util/util.cpp (#8060) * Remove call-command-handler.cpp from src/controller/data_model/gen/ and from src/darwin/Framework/CHIP.xcodeproj/project.pbxproj (#8021) * Add PID to Darwin/Linux Log Lines (#7657) * Add PID to Darwin/Linux Log Lines Problem: - Debugging test failures that involve multiple instances of applications is difficult when the log lines don't indicate which process/application they're coming from. Solution: - Added PID to the log lines on Darwin & Linux minimally to aid in debugging. Testing: - Ran chip-tool on both platforms. * Restyled by clang-format Co-authored-by: Restyled.io <[email protected]> * Update Darwin framework to use GetConnectedDevice (#8011) * Update Darwin framework to use GetConnectedDevice * address review comments * Add a string circular buffer. (#7475) * Add a string circular buffer. * Update src/lib/support/StringCircularBuffer.cpp Co-authored-by: Boris Zbarsky <[email protected]> * Update src/lib/support/StringCircularBuffer.cpp Co-authored-by: Boris Zbarsky <[email protected]> * Update src/lib/support/StringCircularBuffer.cpp Co-authored-by: Boris Zbarsky <[email protected]> * Resolve comments * Resolve comments * Resolve comments * Resolve comments Co-authored-by: Justin Wood <[email protected]> Co-authored-by: Boris Zbarsky <[email protected]> * [lighting-app] Enable optional Basic Cluster attributes (#8083) * [lighting-app] Enable optional Basic Cluster attributes Enable optional Basic Cluster attributes as requested by Chaintanya, to verify a proper behaviour of the Python controller. * Re-gen code * Add GenerateEventPathList and test (#8014) Summary of Changes: -- Add GenerateEventPathList function in ReadClinet -- Add one event path test and two event path test -- Fix im e2e cirque test in chip_im_responder from PR 7634 since cirque test is not enabled * Have a single loopback transport implementation for tests. (#8045) We have multiple, some with different bells and whistles. Should have one shared thing. Unfortunately, some consumers want to drive things off the loopback (like resending reliable messages) that other consumers can't know about because they are at lower layers of the stack. We just use a virtual hook for that for now. * Don't treat a resolve as done in chip-device-ctrl until we have a CASE connection. (#8012) This ensures that following commands will work instead of racing the CASE connection setup. * Regenerating * Restyled by whitespace * Restyled by google-java-format * Regenerating Co-authored-by: Justin Wood <[email protected]> Co-authored-by: Restyled.io <[email protected]> Co-authored-by: Song Guo <[email protected]> Co-authored-by: C Freeman <[email protected]> Co-authored-by: Vivien Nicolas <[email protected]> Co-authored-by: Damian Królik <[email protected]> Co-authored-by: chrisdecenzo <[email protected]> Co-authored-by: rgoliver <[email protected]> Co-authored-by: Wang Qixiang <[email protected]> Co-authored-by: Kevin Schoedel <[email protected]> Co-authored-by: Yufeng Wang <[email protected]> Co-authored-by: Evgeny Margolis <[email protected]> Co-authored-by: Pankaj Garg <[email protected]> Co-authored-by: krypton36 <[email protected]> Co-authored-by: Timothy Maes <[email protected]> Co-authored-by: CodeChronos <[email protected]> Co-authored-by: Vincent Coubard <[email protected]> Co-authored-by: Austin Hsieh <[email protected]> Co-authored-by: Jerry Johns <[email protected]> Co-authored-by: Zang MingJie <[email protected]> Co-authored-by: Boris Zbarsky <[email protected]> Co-authored-by: yunhanw-google <[email protected]>

* Mdns: separate query responders for advertisers The current setup uses a single query responder for commissionable and operational and all the records are cleared together and added in a large group. This makes it challenging to run commissionable and operational at the same time since they can't be cleared individually and because there are duplicate records between operational and commissionable. This commit changes the code to use one query responder per advertisement type and gets the ResponseSender to loop through the query responders to filter for answers. * Apply suggestions from code review Co-authored-by: chrisdecenzo <[email protected]> * Restyled by clang-format * Fix tests - they were not testing right. * Restyled by clang-format * Update number of responders Represents 5 multi-admin fabrics for operational, commissionable and commissioner. * Whoops - broke a test with my last "minor" change. * Lighten the stack load for the failing test. Co-authored-by: chrisdecenzo <[email protected]> Co-authored-by: Restyled.io <[email protected]>

* add electrical measurement cluster to controller-clusters.zap Only a selection of the optional clusters is enabled. mandatory: - measurement type optional: - total active power - rms voltage - rms voltage min - rms voltage max - rms current - rms current min - rms current max - active power - active power min - active power max * add electrical measurement cluster to zap_cluster_list.py * add electrical measurement cluster to all-clusters-app * add electrical measurement cluster to all-cluster-app/esp32 * add app helpers for electrical measurement cluster * fix all cluster app build * regen all * Regenerating * Restyled by whitespace * Restyled by clang-format * Regenerating * Restyled by whitespace * Restyled by google-java-format * [fix] Wrong callback passed to WriteAttribute in Python[TE4] (project-chip#8019) * Fix callback typr * Run codegen * Mdns: _chip* -> _matter*, remove 0's on subtypes (project-chip#7805) * Mdns: _chip* i-> _matter*, remove 0's on subtypes The second ballot changed the format of the subtypes as well as fixed the service name prefix from _chip* to _matter*. NOTE FOR DEVs: Updating past this change will require updating the devices and controllers as otherwise they will be unable to find each other. * Darwin - use local kLocalDomain var. It includes a . and the one in the ServiceNaming doesn't. * Trying again for Darwin. * Fix host name. * Fix small error on merge * Use a dedicated chip-types.xml file into src/app/zap-templates/zcl (project-chip#7873) * Update chip-types.xml * Update gen/ folders * Revert "Mark UDP socket as ready for write in send msg. (project-chip#7927)" (project-chip#8027) Turns out the code causes that subsequent select()-s never wait so another solution for the original issue must be prepared. This reverts commit 8c55523. * [ZAP] Add isAnalog instead of isDiscreteType to account for string types (project-chip#7821) * Mdns: separate query responders for advertisers (project-chip#7615) * Mdns: separate query responders for advertisers The current setup uses a single query responder for commissionable and operational and all the records are cleared together and added in a large group. This makes it challenging to run commissionable and operational at the same time since they can't be cleared individually and because there are duplicate records between operational and commissionable. This commit changes the code to use one query responder per advertisement type and gets the ResponseSender to loop through the query responders to filter for answers. * Apply suggestions from code review Co-authored-by: chrisdecenzo <[email protected]> * Restyled by clang-format * Fix tests - they were not testing right. * Restyled by clang-format * Update number of responders Represents 5 multi-admin fabrics for operational, commissionable and commissioner. * Whoops - broke a test with my last "minor" change. * Lighten the stack load for the failing test. Co-authored-by: chrisdecenzo <[email protected]> Co-authored-by: Restyled.io <[email protected]> * [nrfconnect] Stop using cooperative threads (project-chip#8025) nRF Connect platform used cooperative threads, that is threads that cannot be preempted, to workaround some synchronization issues in the past, but it's no longer needed and is risky. The latter was revealed by project-chip#7927, which caused that `select()` never waits (needs to be fixed in another PR). Signed-off-by: Damian Krolik <[email protected]> * IPV6-only test app for esp32 (examples/ipv6only-app) (project-chip#7289) Adds RPCs that conform to NDM/GDM WiFi capabilities. Disables IPV4 DHCP client. Assigns link-local IPV6 address. Starts a UDP echo server on an IPV6 UDP socket. Test: Ran on M5Stack and verified it scanned and connected to wifi, after connected verified it echoed UDP messages. * [ESP32] fix compile error for persistent-storage app (project-chip#7940) * [ESP32] fix compile error for persistent-storage app * add CI for all ESP examples * Consolidate error types (project-chip#7916) * Consolidate error types #### Problem CHIP currently has `CHIP_ERROR`, `INET_ERROR`, `BLE_ERROR`, `ASN1_ERROR`, `System::Error`. Since they're all integers underneath, there is no type checking, and existing code routinely treats them as interchangeable. There is also existing code that uses `int` and `int32_t`. Each also has a `…_NO_ERROR` code which is independently configurable, but existing code assumes they are equal, and there is exising code that assumes the C convention that the value is 0. #### Change overview Fixes project-chip#7688 Unify chip errors This change is largely a mechanical replacement of type and some error code names, with a few cleanups of weak typing. Consolidated error types to one, `CHIP_ERROR`. Removed configurability of `CHIP_NO_ERROR`; it must be 0. Added `CHIP_ERROR_FORMAT` to replace hardcoded `PRI…32` in `printf()` format strings. Added `CHIP_ERROR_SENTINEL` to replace the use of `CHIP_ERROR_MAX` as a sentinel. Renamed or consolidated the following errors, which have the same meaning. Merely similar codes are left alone. This change contains compatibility defintitions to avoid breaking PRs in flight, which will be removed in a followup. | before | after | |:---------------------------------------|:------------------------------------| | ASN1_ERROR | CHIP_ERROR | | ASN1_NO_ERROR | CHIP_NO_ERROR | | BLE_ERROR | CHIP_ERROR | | BLE_NO_ERROR | CHIP_NO_ERROR | | BLE_ERROR_BAD_ARGS | CHIP_ERROR_INVALID_ARGUMENT | | BLE_ERROR_INCORRECT_STATE | CHIP_ERROR_INCORRECT_STATE | | BLE_ERROR_MESSAGE_INCOMPLETE | CHIP_ERROR_MESSAGE_INCOMPLETE | | BLE_ERROR_NOT_IMPLEMENTED | CHIP_ERROR_NOT_IMPLEMENTED | | BLE_ERROR_NO_ENDPOINTS | CHIP_ERROR_ENDPOINT_POOL_FULL | | BLE_ERROR_NO_MEMORY | CHIP_ERROR_NO_MEMORY | | BLE_ERROR_OUTBOUND_MESSAGE_TOO_BIG | CHIP_ERROR_OUTBOUND_MESSAGE_TOO_BIG | | BLE_ERROR_RECEIVED_MESSAGE_TOO_BIG | CHIP_ERROR_INBOUND_MESSAGE_TOO_BIG | | CHIP_SYSTEM_NO_ERROR | CHIP_NO_ERROR | | CHIP_SYSTEM_ERROR_ACCESS_DENIED | CHIP_ERROR_ACCESS_DENIED | | CHIP_SYSTEM_ERROR_BAD_ARGS | CHIP_ERROR_INVALID_ARGUMENT | | CHIP_SYSTEM_ERROR_NOT_SUPPORTED | CHIP_ERROR_UNSUPPORTED_CHIP_FEATURE | | CHIP_SYSTEM_ERROR_NO_MEMORY | CHIP_ERROR_NO_MEMORY | | CHIP_SYSTEM_ERROR_REAL_TIME_NOT_SYNCED | CHIP_ERROR_REAL_TIME_NOT_SYNCED | | CHIP_SYSTEM_ERROR_UNEXPECTED_EVENT | CHIP_ERROR_UNEXPECTED_EVENT | | CHIP_SYSTEM_ERROR_UNEXPECTED_STATE | CHIP_ERROR_INCORRECT_STATE | | INET_ERROR | CHIP_ERROR | | INET_NO_ERROR | CHIP_NO_ERROR | | INET_ERROR_BAD_ARGS | CHIP_ERROR_INVALID_ARGUMENT | | INET_ERROR_INBOUND_MESSAGE_TOO_BIG | CHIP_ERROR_INBOUND_MESSAGE_TOO_BIG | | INET_ERROR_INCORRECT_STATE | CHIP_ERROR_INCORRECT_STATE | | INET_ERROR_MESSAGE_TOO_LONG | CHIP_ERROR_MESSAGE_TOO_LONG | | INET_ERROR_NO_CONNECTION_HANDLER | CHIP_ERROR_NO_CONNECTION_HANDLER | | INET_ERROR_NO_ENDPOINTS | CHIP_ERROR_ENDPOINT_POOL_FULL | | INET_ERROR_NOT_IMPLEMENTED | CHIP_ERROR_NOT_IMPLEMENTED | | INET_ERROR_NOT_SUPPORTED | CHIP_ERROR_UNSUPPORTED_CHIP_FEATURE | | INET_ERROR_NO_MEMORY | CHIP_ERROR_NO_MEMORY | | INET_ERROR_CONNECTION_ABORTED | CHIP_ERROR_CONNECTION_ABORTED | #### Testing Modified some unit tests to account for consolidated error codes. Built locally with `CHIP_ERROR` hardcoded to `int32_t`, `uint32_t`, `int16_t`, and `uint16_t` to shake out mistaken type assumptions. * Build fixes * Darwin * Doxygen * Fixes * Change default CHIP_CONFIG_ERROR_TYPE back to `int32_t`. * Fix CHIP_ERROR_INBOUND_MESSAGE_TOO_BIG message text. * C++-style casts. * fix: CHIPError.h uses stdint.h * zap * fix for Mbed merge * Throw exception instead of return 0 for get-fabricid in case of error (project-chip#7884) * Implement User Directed Commissioning - Phase 1 (project-chip#8034) * Updated CHIP Certs Implementation to Use Spans instaed of Pointer/Len Pairs. (project-chip#7599) Specifically, added the following FixedSpan types: - CertificateKeyId - P256ECDSASignatureSpan - P256PublicKeySpan Also redefined mString member in the ChipRDN structure to be ByteSpan. * Zap use const expr for clusters commands attributes (project-chip#7946) * Remove duplicated command ids and attribute ids definitions from the encoder-src.zapt template Add ids/Commands.zapt template Update template to use Commands.h and Clusters.h * Fix XML definitions typos for attributes name * Add ids/Commands.h to the gen folder of src/app/common * Update log messages for session expiry (project-chip#8017) * Update log messages for session expiry * update format specifier * Remove client-command-macro.h-aa80d825 from examples/bridge-app/bridge-common/gen (project-chip#8026) * Fix Ble endpoint pool FULL on macOS controller (project-chip#7806) * * Move naming qpg6100 to qpg (project-chip#7936) * Removal Env overrides sdk root and target IC * fix build error in TCP.cpp (project-chip#8072) Include <limits> to satisfy the proper use of std::numeric_limits. * Add lighting example for Mbed OS (project-chip#8048) * Add back structs, bitmaps and enums of the Messaging cluster into ami… (project-chip#8061) * Add back structs, bitmaps and enums of the Messaging cluster into ami.xml * Update gen/ folders * Deprecated ASN1Writer::Finalize() Method. (project-chip#8056) * Split KeyValueStoreManager.java into interface and impl, and add Set() in JNI. (project-chip#8046) * [Test] Add Test_TC_DM_3_1 (project-chip#8067) * Add Test_TC_DM_3_1 to src/app/tests/suites with all tests disabled since the attributes are not present into commissioning.xml * Update generated tests * Followup project-chip#7873 - Replace emberAfPutInt8uInResp(ZCL_DEFAULT_RESPONSE_COMMAND_ID) by emberAfPutInt32uInResp(ZCL_DEFAULT_RESPONSE_COMMAND_ID) in src/app/util/util.cpp (project-chip#8060) * Remove call-command-handler.cpp from src/controller/data_model/gen/ and from src/darwin/Framework/CHIP.xcodeproj/project.pbxproj (project-chip#8021) * Add PID to Darwin/Linux Log Lines (project-chip#7657) * Add PID to Darwin/Linux Log Lines Problem: - Debugging test failures that involve multiple instances of applications is difficult when the log lines don't indicate which process/application they're coming from. Solution: - Added PID to the log lines on Darwin & Linux minimally to aid in debugging. Testing: - Ran chip-tool on both platforms. * Restyled by clang-format Co-authored-by: Restyled.io <[email protected]> * Update Darwin framework to use GetConnectedDevice (project-chip#8011) * Update Darwin framework to use GetConnectedDevice * address review comments * Add a string circular buffer. (project-chip#7475) * Add a string circular buffer. * Update src/lib/support/StringCircularBuffer.cpp Co-authored-by: Boris Zbarsky <[email protected]> * Update src/lib/support/StringCircularBuffer.cpp Co-authored-by: Boris Zbarsky <[email protected]> * Update src/lib/support/StringCircularBuffer.cpp Co-authored-by: Boris Zbarsky <[email protected]> * Resolve comments * Resolve comments * Resolve comments * Resolve comments Co-authored-by: Justin Wood <[email protected]> Co-authored-by: Boris Zbarsky <[email protected]> * [lighting-app] Enable optional Basic Cluster attributes (project-chip#8083) * [lighting-app] Enable optional Basic Cluster attributes Enable optional Basic Cluster attributes as requested by Chaintanya, to verify a proper behaviour of the Python controller. * Re-gen code * Add GenerateEventPathList and test (project-chip#8014) Summary of Changes: -- Add GenerateEventPathList function in ReadClinet -- Add one event path test and two event path test -- Fix im e2e cirque test in chip_im_responder from PR 7634 since cirque test is not enabled * Have a single loopback transport implementation for tests. (project-chip#8045) We have multiple, some with different bells and whistles. Should have one shared thing. Unfortunately, some consumers want to drive things off the loopback (like resending reliable messages) that other consumers can't know about because they are at lower layers of the stack. We just use a virtual hook for that for now. * Don't treat a resolve as done in chip-device-ctrl until we have a CASE connection. (project-chip#8012) This ensures that following commands will work instead of racing the CASE connection setup. * Regenerating * Restyled by whitespace * Restyled by google-java-format * Regenerating Co-authored-by: Justin Wood <[email protected]> Co-authored-by: Restyled.io <[email protected]> Co-authored-by: Song Guo <[email protected]> Co-authored-by: C Freeman <[email protected]> Co-authored-by: Vivien Nicolas <[email protected]> Co-authored-by: Damian Królik <[email protected]> Co-authored-by: chrisdecenzo <[email protected]> Co-authored-by: rgoliver <[email protected]> Co-authored-by: Wang Qixiang <[email protected]> Co-authored-by: Kevin Schoedel <[email protected]> Co-authored-by: Yufeng Wang <[email protected]> Co-authored-by: Evgeny Margolis <[email protected]> Co-authored-by: Pankaj Garg <[email protected]> Co-authored-by: krypton36 <[email protected]> Co-authored-by: Timothy Maes <[email protected]> Co-authored-by: CodeChronos <[email protected]> Co-authored-by: Vincent Coubard <[email protected]> Co-authored-by: Austin Hsieh <[email protected]> Co-authored-by: Jerry Johns <[email protected]> Co-authored-by: Zang MingJie <[email protected]> Co-authored-by: Boris Zbarsky <[email protected]> Co-authored-by: yunhanw-google <[email protected]>

boring-cyborg bot added the lib label Jun 14, 2021

pullapprove bot requested review from andy31415, bzbarsky-apple, chrisdecenzo, Damian-Nordic, hawk248, jepenven-silabs and msandstedt June 14, 2021 18:40

pullapprove bot added the review - pending label Jun 14, 2021

cecille requested a review from sweetymhaiske June 14, 2021 20:15

cecille force-pushed the mdns_multiple_query_responders branch from 5e33f3d to e1f1164 Compare June 14, 2021 23:15

chrisdecenzo requested changes Jun 15, 2021

View reviewed changes

restyled-io bot mentioned this pull request Jun 16, 2021

Restyle Mdns: separate query responders for advertisers #7670

Closed

cecille and others added 2 commits June 17, 2021 01:59

Apply suggestions from code review

924facd

Co-authored-by: chrisdecenzo <[email protected]>

woody-apple force-pushed the mdns_multiple_query_responders branch from 08abc00 to 924facd Compare June 17, 2021 01:59

andy31415 requested a review from chrisdecenzo June 18, 2021 21:27

restyled-commits and others added 2 commits June 21, 2021 12:04

Restyled by clang-format

e871da8

Merge remote-tracking branch 'upstream/master' into mdns_multiple_que…

82ee47e

…ry_responders

stale bot added stale Stale issue or PR and removed stale Stale issue or PR labels Jun 28, 2021

restyled-io bot mentioned this pull request Jun 29, 2021

Restyle Mdns: separate query responders for advertisers #8002

Closed

cecille force-pushed the mdns_multiple_query_responders branch from 51d87c5 to c90483f Compare June 29, 2021 14:57

cecille force-pushed the mdns_multiple_query_responders branch from c90483f to 637c317 Compare June 29, 2021 18:05

Fix tests - they were not testing right.

ab1b0d1

cecille force-pushed the mdns_multiple_query_responders branch from 637c317 to ab1b0d1 Compare June 29, 2021 18:09

woody-apple approved these changes Jun 29, 2021

View reviewed changes

restyled-commits and others added 2 commits June 29, 2021 16:28

Restyled by clang-format

fd119f6

Update number of responders

e74cfad

Represents 5 multi-admin fabrics for operational, commissionable and commissioner.

msandstedt approved these changes Jun 30, 2021

View reviewed changes

Whoops - broke a test with my last "minor" change.

b230276

cecille added 2 commits June 30, 2021 10:16

Lighten the stack load for the failing test.

ad4fce3

Merge remote-tracking branch 'upstream/master' into mdns_multiple_que…

b83cbfc

…ry_responders

chrisdecenzo approved these changes Jun 30, 2021

View reviewed changes

pullapprove bot added review - approved and removed review - pending labels Jun 30, 2021

woody-apple merged commit f6f32fa into project-chip:master Jun 30, 2021

cecille deleted the mdns_multiple_query_responders branch July 23, 2021 19:34

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Mdns: separate query responders for advertisers #7615

Mdns: separate query responders for advertisers #7615

cecille commented Jun 14, 2021

todo bot commented Jun 14, 2021

todo bot commented Jun 14, 2021

cecille commented Jun 14, 2021

chrisdecenzo left a comment

woody-apple commented Jun 16, 2021

woody-apple commented Jun 17, 2021

woody-apple commented Jun 21, 2021

stale bot commented Jun 28, 2021

woody-apple commented Jun 29, 2021

todo bot commented Jun 29, 2021

todo bot commented Jun 30, 2021

Mdns: separate query responders for advertisers #7615

Mdns: separate query responders for advertisers #7615

Conversation

cecille commented Jun 14, 2021

Problem

Change overview

Testing

todo bot commented Jun 14, 2021

When multi-admin is enabled, commissionable does not need to be cleared here.

This comment was generated by todo based on a TODO comment in 5e33f3d in #7615. cc @cecille.

todo bot commented Jun 14, 2021

When multi-admin is enabled, operational does not need to be cleared here.

This comment was generated by todo based on a TODO comment in 5e33f3d in #7615. cc @cecille.

cecille commented Jun 14, 2021

chrisdecenzo left a comment

Choose a reason for hiding this comment

woody-apple commented Jun 16, 2021

woody-apple commented Jun 17, 2021

woody-apple commented Jun 21, 2021

stale bot commented Jun 28, 2021

woody-apple commented Jun 29, 2021

todo bot commented Jun 29, 2021

(cecille): Template this and set appropriately. Please see issue #8000.

This comment was generated by todo based on a TODO comment in e74cfad in #7615. cc @cecille.

todo bot commented Jun 30, 2021

(cecille): Fix this test once #8000 gets resolved.

This comment was generated by todo based on a TODO comment in b230276 in #7615. cc @cecille.

This comment was generated by todo based on a `TODO` comment in `5e33f3d` in #7615. cc @cecille.

This comment was generated by todo based on a `TODO` comment in `5e33f3d` in #7615. cc @cecille.

This comment was generated by todo based on a `TODO` comment in `e74cfad` in #7615. cc @cecille.

This comment was generated by todo based on a `TODO` comment in `b230276` in #7615. cc @cecille.