[trace] introduce trace failed to Ext #11019

ordian · 2019-09-03T12:51:18Z

When a subcall fails, we need to clean up the tracing data (last_mem_written, last_store_written) in the ExecutiveVMTracer. Also to properly track subcalls, we maintain a stack of tracing data.

One approach of doing that is to pass mem_written directly to trace_executed. It does introduce a perf regression.

This PR adds a method trace_failed to Ext, which is called when the execution of an instruction has failed. An alternative suggested by @tomusdrw would be to merge trace_failed into trace_executed by passing a Result<(stack_push: &[U256], mem: &[u8]), ()>, but I'm not sure if is any better.

* master: Extract snapshot to own crate (#11010) Edit publish-onchain.sh to use https (#11016)

tomusdrw

I was suggesting to merge the two methods to simplify assumptions for implementation (i.e. trace_executed is ALWAYS called after trace_prepare_execute). It would also deduplicate poping from the stack in the implementation of VMTracer.

But looks good anyway!

tomusdrw · 2019-09-03T20:29:58Z

ethcore/vm/src/ext.rs

@@ -166,6 +166,9 @@ pub trait Ext {
 	/// Prepare to trace an operation. Passthrough for the VM trace.
 	fn trace_prepare_execute(&mut self, _pc: usize, _instruction: u8, _gas_cost: U256, _mem_written: Option<(usize, usize)>, _store_written: Option<(U256, U256)>) {}


Can we improve the docs here and underline the fact that this call is ALWAYS followed by either trace_executed or trace_failed. and that the implementation is responsible to clean up in either/both of those methods.

dvdplm

Why did you opt to remove the bounds checks here? For performance reasons? Unless it's a severe regression I think it'd be good to have.

ordian · 2019-09-05T09:35:31Z

Why did you opt to remove the bounds checks here? For performance reasons? Unless it's a severe regression I think it'd be good to have.

If the bounds could still be incorrect, it would indicate another bug and a warning could easily be missed/ignored. But if we reach a consensus on that the checks should be added back, I don't mind.

dvdplm · 2019-09-05T09:37:51Z

But if we reach a consensus on that the checks should be added back, I don't mind.

@tomusdrw thoughts? Better to crash?

tomusdrw · 2019-09-05T10:50:35Z

@dvdplm I'd rather bound the o and o+s with mem.len() to be safe. The bound checking and a warning was introducing a lot of noise.
To be fair I'd rather not crash and have potentially incorrect data in tracing.

dvdplm · 2019-09-05T13:21:33Z

f818b23 lgtm

* master: fix: remove unused error-chain (#11028) fix: remove needless use of itertools (#11029) Convert `std::test` benchmarks to use Criterion (#10999) Fix block detail updating (#11015) [trace] introduce trace failed to Ext (#11019) cli: update usage and version headers (#10924) [private-tx] remove unused rand (#11024)

…he-right-place * master: cleanup json crate (#11027) [spec] add istanbul test spec (#11033) [json-spec] make blake2 pricing spec more readable (#11034) Add blake2_f precompile (#11017) Add new line after writing block to hex file. (#10984) fix: remove unused error-chain (#11028) fix: remove needless use of itertools (#11029) Convert `std::test` benchmarks to use Criterion (#10999) Fix block detail updating (#11015) [trace] introduce trace failed to Ext (#11019) cli: update usage and version headers (#10924) [private-tx] remove unused rand (#11024)

* [trace] add trace_failed to Ext, manage stack of trace data * [evm] call trace_failed only if self.do_trace * [evm] add a comment about do_trace set to true * [vm] improve the doc in trace_prepare_execute * [trace] add the bounds check back

This reverts commit 895dc54.

* [trace] add trace_failed to Ext, manage stack of trace data * [evm] call trace_failed only if self.do_trace * [evm] add a comment about do_trace set to true * [vm] improve the doc in trace_prepare_execute * [trace] add the bounds check back

* add more tx tests (#11038) * Fix parallel transactions race-condition (#10995) * Add blake2_f precompile (#11017) * [trace] introduce trace failed to Ext (#11019) * Edit publish-onchain.sh to use https (#11016) * Fix deadlock in network-devp2p (#11013) * EIP 1108: Reduce alt_bn128 precompile gas costs (#11008) * xDai chain support and nodes list update (#10989) * EIP 2028: transaction gas lowered from 68 to 16 (#10987) * EIP-1344 Add CHAINID op-code (#10983) * manual publish jobs for releases, no changes for nightlies (#10977) * [blooms-db] Fix benchmarks (#10974) * Verify transaction against its block during import (#10954) * Better error message for rpc gas price errors (#10931) * Fix fork choice (#10837) * Fix compilation on recent nightlies (#10991)

* add more tx tests (#11038) * Fix parallel transactions race-condition (#10995) * Add blake2_f precompile (#11017) * [trace] introduce trace failed to Ext (#11019) * Edit publish-onchain.sh to use https (#11016) * Fix deadlock in network-devp2p (#11013) * EIP 1108: Reduce alt_bn128 precompile gas costs (#11008) * xDai chain support and nodes list update (#10989) * EIP 2028: transaction gas lowered from 68 to 16 (#10987) * EIP-1344 Add CHAINID op-code (#10983) * manual publish jobs for releases, no changes for nightlies (#10977) * [blooms-db] Fix benchmarks (#10974) * Verify transaction against its block during import (#10954) * Better error message for rpc gas price errors (#10931) * tx-pool: accept local tx with higher gas price when pool full (#10901) * Fix fork choice (#10837) * Cleanup unused vm dependencies (#10787) * Fix compilation on recent nightlies (#10991)

* master: (70 commits) ethcore: remove `test-helper feat` from build (#11047) Include test-helpers from ethjson (#11045) [ethcore]: cleanup dependencies (#11043) add more tx tests (#11038) Fix parallel transactions race-condition (#10995) [ethcore]: make it compile without `test-helpers` feature (#11036) Benchmarks for block verification (#11035) Move snapshot related traits to their proper place (#11012) cleanup json crate (#11027) [spec] add istanbul test spec (#11033) [json-spec] make blake2 pricing spec more readable (#11034) Add blake2_f precompile (#11017) Add new line after writing block to hex file. (#10984) fix: remove unused error-chain (#11028) fix: remove needless use of itertools (#11029) Convert `std::test` benchmarks to use Criterion (#10999) Fix block detail updating (#11015) [trace] introduce trace failed to Ext (#11019) cli: update usage and version headers (#10924) [private-tx] remove unused rand (#11024) ...

ordian added 3 commits September 2, 2019 16:51

[trace] add trace_failed to Ext, manage stack of trace data

bfc7d1b

[evm] call trace_failed only if self.do_trace

de1efb2

Merge branch 'master' into ao-introduce-trace-failed

17e2156

* master: Extract snapshot to own crate (#11010) Edit publish-onchain.sh to use https (#11016)

ordian added A3-inprogress ⏳ Pull request is in progress. No review needed at this stage. B1-patch-beta 🕷🕷 M4-core ⛓ Core client code / Rust. B0-patch-stable 🕷 Pull request should also be back-ported to the stable branch. labels Sep 3, 2019

ordian added this to the 2.7 milestone Sep 3, 2019

ordian requested review from sorpaas and tomusdrw September 3, 2019 12:53

[evm] add a comment about do_trace set to true

020b324

tomusdrw approved these changes Sep 3, 2019

View reviewed changes

[vm] improve the doc in trace_prepare_execute

8196c62

ordian marked this pull request as ready for review September 4, 2019 13:33

ordian added A0-pleasereview 🤓 Pull request needs code review. and removed A3-inprogress ⏳ Pull request is in progress. No review needed at this stage. labels Sep 4, 2019

dvdplm approved these changes Sep 5, 2019

View reviewed changes

[trace] add the bounds check back

f818b23

dvdplm merged commit 44c00b1 into master Sep 5, 2019

ordian deleted the ao-introduce-trace-failed branch September 5, 2019 14:18

s3krit added a commit that referenced this pull request Sep 11, 2019

Revert "[trace] introduce trace failed to Ext (#11019)"

5c95e70

This reverts commit 895dc54.

This was referenced Sep 12, 2019

v2.6.3-beta #11040

Merged

v2.5.8-stable #11041

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[trace] introduce trace failed to Ext #11019

[trace] introduce trace failed to Ext #11019

ordian commented Sep 3, 2019 •

edited

Loading

tomusdrw left a comment

tomusdrw Sep 3, 2019

dvdplm left a comment

ordian commented Sep 5, 2019

dvdplm commented Sep 5, 2019 •

edited

Loading

tomusdrw commented Sep 5, 2019

dvdplm commented Sep 5, 2019

		@@ -166,6 +166,9 @@ pub trait Ext {
		/// Prepare to trace an operation. Passthrough for the VM trace.
		fn trace_prepare_execute(&mut self, _pc: usize, _instruction: u8, _gas_cost: U256, _mem_written: Option<(usize, usize)>, _store_written: Option<(U256, U256)>) {}

[trace] introduce trace failed to Ext #11019

[trace] introduce trace failed to Ext #11019

Conversation

ordian commented Sep 3, 2019 • edited Loading

tomusdrw left a comment

Choose a reason for hiding this comment

tomusdrw Sep 3, 2019

Choose a reason for hiding this comment

dvdplm left a comment

Choose a reason for hiding this comment

ordian commented Sep 5, 2019

dvdplm commented Sep 5, 2019 • edited Loading

tomusdrw commented Sep 5, 2019

dvdplm commented Sep 5, 2019

ordian commented Sep 3, 2019 •

edited

Loading

dvdplm commented Sep 5, 2019 •

edited

Loading