You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In unit test System.Net.Http.Functional.Tests.HttpClientHandler_RemoteServerTest.GetAsync_RequestHeadersAddCustomHeaders_HeaderAndEmptyValueSent(remoteServer: (BaseUri: https://127.0.0.1:61813/, HttpVersion: 1.1) on WASM and browser.
System.NullReferenceException : Object reference not set to an instance of an object.
at System.Net.Http.BrowserHttpHandler.SendAsync(HttpRequestMessage request, CancellationToken cancellationToken)
at System.Net.Http.HttpClient.<SendAsync>g__Core|83_0(HttpRequestMessage request, HttpCompletionOption completionOption, CancellationTokenSource cts, Boolean disposeCts, CancellationTokenSource pendingRequestsCts, CancellationToken originalCancellationToken)
at System.Net.Http.Functional.Tests.HttpClientHandler_RemoteServerTest.GetAsync_RequestHeadersAddCustomHeaders_HeaderAndEmptyValueSent(RemoteServer remoteServer, Uri uri)
Looking at the test I don't see an obvious reason it should fail like this. Is this failing on all browser lanes? There was some work done allow trimming more of the header logic out and if these tests weren't running they may have regressed.
This time different test and on ubuntu, not windows. System.Net.Http.Functional.Tests.HttpClientHandler_RemoteServerTest.GetAsync_ServerNeedsAuthAndNoCredential_StatusCodeUnauthorized
changed the title [-][wasm] System.Net.Http.BrowserHttpHandler.SendAsync() throws NPE[/-][+][wasm] System.Net.Http.BrowserHttpHandler.SendAsync() throws NRE[/+]on Jun 10, 2021
System.ArgumentNullException : Value cannot be null. (Parameter 'fetchResponse')
info: at System.Net.Http.BrowserHttpHandler.SendAsync(HttpRequestMessage request, CancellationToken cancellationToken)
Most likely explanation is a GC issue. I think if it were memory corruption in the runtime (which has been known to happen, but hasn't - afaik - been seen recently) it wouldn't just be consistently producing NREs, other bad stuff would be happening.
[20:02:53] info: Initializing.....
[20:02:55] info: {"method":"console.error","payload":{},"arguments":[{}]}
[20:02:55] fail: Error: System.NullReferenceException: Object reference not set to an instance of an object.
at System.Runtime.InteropServices.JavaScript.Runtime.<>c__DisplayClass27_0.<SetupJSContinuation>g__Complete|0()
at System.Runtime.InteropServices.JavaScript.Runtime.SetupJSContinuation(Task task, JSObject continuationObj)
at Object._convert_exception_for_method_call (http://127.0.0.1:35977/dotnet.js:1:171696)
at Object._handle_exception_for_call (http://127.0.0.1:35977/dotnet.js:1:173665)
at Object._handle_exception_and_produce_result_for_call (http://127.0.0.1:35977/dotnet.js:1:173986)
at Object._call_method_with_converted_args (http://127.0.0.1:35977/dotnet.js:1:175060)
at Object.call_method (http://127.0.0.1:35977/dotnet.js:1:173455)
at Object._unbox_task_rooted (http://127.0.0.1:35977/dotnet.js:1:152563)
at Object._unbox_mono_obj_rooted_with_known_nonprimitive_type (http://127.0.0.1:35977/dotnet.js:1:153449)
at Object._unbox_mono_obj_root (http://127.0.0.1:35977/dotnet.js:1:154685)
at Object._handle_exception_and_produce_result_for_call (http://127.0.0.1:35977/dotnet.js:1:174106)
at Object._call_method_with_converted_args (http://127.0.0.1:35977/dotnet.js:1:175060)
[20:02:55] info: Discovering: System.Xml.Linq.xNodeBuilder.Tests.dll (method display = ClassAndMethod, method display options = None)
[20:02:55] info: Discovered: System.Xml.Linq.xNodeBuilder.Tests.dll (found 7 of 8 test cases)
[20:02:55] info: Starting: System.Xml.Linq.xNodeBuilder.Tests.dll
[20:02:55] info: Finished: System.Xml.Linq.xNodeBuilder.Tests.dll
[20:02:55] info:
[20:02:55] info: === TEST EXECUTION SUMMARY ===
Activity
- improve doc
lewing commentedon Jun 8, 2021
Looking at the test I don't see an obvious reason it should fail like this. Is this failing on all browser lanes? There was some work done allow trimming more of the header logic out and if these tests weren't running they may have regressed.
[wasm][testing] hosting echo server in xharness process (#52923)
pavelsavara commentedon Jun 10, 2021
Here is example of the failure on CI
https://helixre8s23ayyeko0k025g8.blob.core.windows.net/dotnet-runtime-refs-pull-53925-merge-160a064b73f24741b2/System.Net.Http.Functional.Tests/console.3a6c0e9f.log?sv=2019-07-07&se=2021-06-30T09%3A14%3A59Z&sr=c&sp=rl&sig=qBO1wWtqkq9AGYRX0b82P6An%2BBE0LKFnuckJY0ed6BU%3D
This time different test and on ubuntu, not windows.
System.Net.Http.Functional.Tests.HttpClientHandler_RemoteServerTest.GetAsync_ServerNeedsAuthAndNoCredential_StatusCodeUnauthorizedMaybe race condition ?
pavelsavara commentedon Jun 10, 2021
Yes second CI run on the same code as above is green.
[-][wasm] System.Net.Http.BrowserHttpHandler.SendAsync() throws NPE[/-][+][wasm] System.Net.Http.BrowserHttpHandler.SendAsync() throws NRE[/+]pavelsavara commentedon Jun 10, 2021
Another locally
pavelsavara commentedon Jun 10, 2021
pavelsavara commentedon Jun 10, 2021
Repro at pavelsavara@a2b0a7b#diff-66ce677ded58c3a6a0639d321f4460bb0b2c4bcaeb3216ff3db4381937365bd8
It fails randomly with
s,q,x3andx10as laststateI think JSObject is returned as null from some of the marshaling calls.
@lewing @kg thoughts ?
kg commentedon Jun 10, 2021
Most likely explanation is a GC issue. I think if it were memory corruption in the runtime (which has been known to happen, but hasn't - afaik - been seen recently) it wouldn't just be consistently producing NREs, other bad stuff would be happening.
pavelsavara commentedon Jun 19, 2021
Very simple steps to reproduce here https://github.com/dotnet/runtime/pull/54453/files
lewing commentedon Jun 28, 2021
looks similar to
from https://dev.azure.com/dnceng/public/_build/results?buildId=1206770&view=logs&j=108d2c4a-8a62-5a58-8dad-8e1042acc93c&t=568f884b-cc12-5fd3-e7fe-790b5ac403f4&l=125