More thorough quick, and a quicker one by ggreif · Pull Request #186 · caffeinelabs/motoko

ggreif · 2019-02-24T09:53:39Z

quicker: these don't run dfinity tests; good for fast, shallow testing

In turn now target quick runs all dfinity dvm tests in parallel, passing individual DB folders to dvm.

Usage:

make quicker
# or
make quick

Suggestions welcome!

these don't run dfinity tests but in turn now target 'quick' runs all dfinity dvm tests in parallel

test/dvm-db.sh

nomeata

See inline comment.

nomeata · 2019-02-27T11:50:38Z

Also:

Can you rename quick to parallel and quickest to quick
Use parallel in default.nix.

there is now 'parallel' which runs all the tests in parallel, and 'quick' which does the same, but without the dfinity tests. Moved tmpdir-related logic into `dvm.sh`.

run parallel tests from nix too

test/dvm.sh

So that one can run make test ASC=asc where `asc` is on the PATH.

nomeata · 2019-02-27T16:02:46Z

I fixed some problems of running it from nix for you.

ggreif · 2019-02-28T21:49:15Z

Somehow Jenkins is still unhappy, local nix-build works flawlessly.

make --no-print-directory --load-average -j -C run-dfinity quick
overflow: [tc] [run] [run-low] [run-ir] [wasm] [dvm]
--- overflow.dvm-run (expected)
+++ overflow.dvm-run (actual)
@@ -1,4 +1,5 @@
 W, hypervisor: call failed with trap message: Uncaught RuntimeError: unreachable
+W, hypervisor: call failed with trap message: Uncaught RuntimeError: unreachable
 Top-level code done.
 This is reachable.
 This is reachable.
Some tests failed.

nomeata

Looks good to me so far.

nomeata · 2019-02-28T22:33:27Z

Somehow Jenkins is still unhappy, local nix-build works flawlessly.

This is indeed odd, and somewhat unrelated to your change. I actually expect two traps to be reported here. I don’t know why only one is reported. Maybe somehow related to https://github.com/dfinity-lab/dev/commit/21a5f27c9a382a1e462e48ffe6971697c2820eb6?
@jwiegley Do you have an idea why, given a call tree with two trapping messages, dvm would sometimes print one and sometimes print two?

nomeata · 2019-02-28T22:52:56Z

To unblock this, Gabor, feel free to clean up the test case a bit to only trap once, possibly by splitting it up into multiple files. In fact, if you split it up in one file per test, we don’t even need to (ab)use async blocks this way.

jwiegley · 2019-03-01T01:12:28Z

@nomeata Sadly I know very little about dvm, perhaps @blynn can help?

FloorLamp · 2019-03-01T06:00:15Z

I see it trapping twice:

T, hypervisor/sys: runFunction: 1.func$62(6)
T, hypervisor/sys: runFunction: actor hash: ce455108efe743938d380dd05563a347421381c0
T, hypervisor/sys: runFunction: persisted globals = fromList [...]
T, hypervisor/sys: runFunction: globals refs = [(0,WI32 93),(1,WI32 94)]
T, hypervisor/memory: persistent memory: actor is transient
T, hypervisor/sys: invoking module function func$62
D, hypervisor/V8 creating store
D, hypervisor/V8 stack base at store creation time = 0x00007ffee468b4bc
D, hypervisor/V8 creating instance
D, hypervisor/V8 creating module
D, hypervisor/V8 initializing handle
D, hypervisor/V8 registering cleanup actions
D, hypervisor/V8 module instantiated
D, hypervisor/V8 resolving imports and creating instance
...
D, hypervisor/V8 determining module exports
D, hypervisor/V8 retrieving function exports
D, hypervisor/V8 retrieving global exports
D, hypervisor/V8 setting global 0 to WI32 93
D, hypervisor/V8 setting global 1 to WI32 94
D, hypervisor/V8 > getFunctionByName(name: func$62)
D, hypervisor/V8 > type (i32) -> ()
D, hypervisor/V8 > invoke(args: [WI32 6])
D, hypervisor/V8 stack base at invocation time = 0x00007ffee468b4bc
D, hypervisor/V8 expecting args array of size 1
D, hypervisor/V8 allocating result array of size 0
T, hypervisor/sys: data.length(ref: 93, value: ...
T, hypervisor/sys: data.internalize(dst: 4096, len: 4992, src: 93, offset: 0) = value: ...
T, hypervisor/sys: elem.length(src: 94) = length: 12, value: ...
T, hypervisor/sys: elem.internalize(dst: 0, len: 12, src: 94, offset: 0) = value: ...
D, hypervisor/V8 function call FAILED
D, hypervisor/V8 > globals() = globals: [(0,WI32 93),(1,WI32 94)]
D, hypervisor/V8 maybe deleting wasm_table_t
D, hypervisor/V8 maybe deleting wasm_memory_t
D, hypervisor/V8 deleting exports
D, hypervisor/V8 deleting imports
D, hypervisor/V8 deleting wasm_instance_t
D, hypervisor/V8 deleting wasm_module_t
D, hypervisor/V8 deleting store
W, hypervisor: call failed with trap message: Uncaught RuntimeError: unreachable
T, hypervisor/memory: persistent memory: no memory working copy to remove
T, hypervisor/sys: runFunction: 1.func$68(8)
T, hypervisor/sys: runFunction: actor hash: ce455108efe743938d380dd05563a347421381c0
T, hypervisor/sys: runFunction: persisted globals = fromList [...]
T, hypervisor/sys: runFunction: globals refs = [(0,WI32 93),(1,WI32 94)]
T, hypervisor/memory: persistent memory: actor is transient
T, hypervisor/sys: invoking module function func$68
D, hypervisor/V8 creating store
D, hypervisor/V8 stack base at store creation time = 0x00007ffee468b4bc
D, hypervisor/V8 creating instance
D, hypervisor/V8 creating module
D, hypervisor/V8 initializing handle
D, hypervisor/V8 registering cleanup actions
D, hypervisor/V8 module instantiated
D, hypervisor/V8 resolving imports and creating instance
...
D, hypervisor/V8 determining module exports
D, hypervisor/V8 retrieving function exports
D, hypervisor/V8 retrieving global exports
D, hypervisor/V8 setting global 0 to WI32 93
D, hypervisor/V8 setting global 1 to WI32 94
D, hypervisor/V8 > getFunctionByName(name: func$68)
D, hypervisor/V8 > type (i32) -> ()
D, hypervisor/V8 > invoke(args: [WI32 8])
D, hypervisor/V8 stack base at invocation time = 0x00007ffee468b4bc
D, hypervisor/V8 expecting args array of size 1
D, hypervisor/V8 allocating result array of size 0
T, hypervisor/sys: data.length(ref: 93, value: ...
T, hypervisor/sys: data.internalize(dst: 4096, len: 4992, src: 93, offset: 0) = value: ...
T, hypervisor/sys: elem.length(src: 94) = length: 12, value: ...
T, hypervisor/sys: elem.internalize(dst: 0, len: 12, src: 94, offset: 0) = value: ...
D, hypervisor/V8 function call FAILED
D, hypervisor/V8 > globals() = globals: [(0,WI32 93),(1,WI32 94)]
D, hypervisor/V8 maybe deleting wasm_table_t
D, hypervisor/V8 maybe deleting wasm_memory_t
D, hypervisor/V8 deleting exports
D, hypervisor/V8 deleting imports
D, hypervisor/V8 deleting wasm_instance_t
D, hypervisor/V8 deleting wasm_module_t
D, hypervisor/V8 deleting store
W, hypervisor: call failed with trap message: Uncaught RuntimeError: unreachable
T, hypervisor/memory: persistent memory: no memory working copy to remove

ggreif · 2019-03-01T23:51:04Z

@nomeata the obvious difference is that we now have potentially several dvms running on the same host. Is there a possibility of interference (shared system resources, files, etc.)?

nomeata · 2019-03-02T09:45:08Z

@nomeata the obvious difference is that we now have potentially several dvms running on the same host. Is there a possibility of interference (shared system resources, files, etc.)?

I sincerely hope that with different --db parameters, that they would not interfere.

ggreif · 2019-03-09T15:48:49Z

default.nix

    '' +
      (if test-dvm then ''
-      make -C test/run-dfinity VERBOSE=1 ASC=asc quick
+      make --load-average -j -C test/run-dfinity VERBOSE=1 ASC=asc quick


without this, there was no failure, so there is some kind of crosstalk between VMs :-(

This should be investigated seriously.

## Changelog for ic-hs: Branch: master Commits: [dfinity/ic-hs@9152a0ff...189b3a3c](dfinity/ic-hs@9152a0f...189b3a3) * [`c88bb6c7`](dfinity/ic-hs@c88bb6c) trim long blobs in json debug state ([dfinity/ic-hs⁠#192](https://github.com/dfinity/ic-hs/issues/192)) * [`189b3a3c`](dfinity/ic-hs@189b3a3) implement IC-1462 ([dfinity/ic-hs⁠#186](https://github.com/dfinity/ic-hs/issues/186))

## Changelog for ic-hs: Branch: master Commits: [dfinity/ic-hs@9152a0ff...46287711](dfinity/ic-hs@9152a0f...4628771) * [`c88bb6c7`](dfinity/ic-hs@c88bb6c) trim long blobs in json debug state ([dfinity/ic-hs⁠#192](https://github.com/dfinity/ic-hs/issues/192)) * [`189b3a3c`](dfinity/ic-hs@189b3a3) implement IC-1462 ([dfinity/ic-hs⁠#186](https://github.com/dfinity/ic-hs/issues/186)) * [`46287711`](dfinity/ic-hs@4628771) bump canister version upon successful message execution and tighten tests ([dfinity/ic-hs⁠#187](https://github.com/dfinity/ic-hs/issues/187))

add targets 'quicker'

7478424

these don't run dfinity tests but in turn now target 'quick' runs all dfinity dvm tests in parallel

ggreif requested review from FloorLamp and nomeata February 24, 2019 09:53

tweak

4999208

nomeata suggested changes Feb 24, 2019

View reviewed changes

test/dvm-db.sh Outdated Show resolved Hide resolved

nomeata suggested changes Feb 24, 2019

View reviewed changes

ggreif added 2 commits February 27, 2019 15:02

review feedback

bc6018f

there is now 'parallel' which runs all the tests in parallel, and 'quick' which does the same, but without the dfinity tests. Moved tmpdir-related logic into `dvm.sh`.

review feedback

2a14343

run parallel tests from nix too

nomeata reviewed Feb 27, 2019

View reviewed changes

test/dvm.sh Outdated Show resolved Hide resolved

nomeata added 2 commits February 27, 2019 17:00

Make $(ASC) dependency of _out/%.done optional

1308bad

So that one can run make test ASC=asc where `asc` is on the PATH.

Use mkdir -p

bd2e7a9

ggreif added 3 commits February 28, 2019 21:12

use order-only prereq for _out

e622174

cheap cleanup

b7323fc

use mktemp to create and trap to clean up

09e7162

nomeata approved these changes Feb 28, 2019

View reviewed changes

ggreif added 4 commits March 9, 2019 00:39

Merge branch 'master' into gabor/parallel-test

4c1a923

update output

b7eec03

test an idea

59a9768

run run-dfinity in parallel

645e51b

ggreif commented Mar 9, 2019

View reviewed changes

just 2 jobs

16551b6

ggreif added 2 commits March 9, 2019 16:53

test 4 jobs

26158f4

try 8 jobs

96b5707

ggreif marked this pull request as ready for review March 9, 2019 15:57

ggreif merged commit 16f26cc into master Mar 9, 2019

Conversation

ggreif commented Feb 24, 2019

Uh oh!

Uh oh!

nomeata left a comment

Choose a reason for hiding this comment

Uh oh!

nomeata commented Feb 27, 2019

Uh oh!

Uh oh!

nomeata commented Feb 27, 2019

Uh oh!

ggreif commented Feb 28, 2019

Uh oh!

nomeata left a comment

Choose a reason for hiding this comment

Uh oh!

nomeata commented Feb 28, 2019

Uh oh!

nomeata commented Feb 28, 2019

Uh oh!

jwiegley commented Mar 1, 2019

Uh oh!

FloorLamp commented Mar 1, 2019

Uh oh!

ggreif commented Mar 1, 2019

Uh oh!

nomeata commented Mar 2, 2019

Uh oh!

ggreif Mar 9, 2019

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants