when use trace_processor_shell load long trace file , process is killed and echos "Killedg trace: 1248.58 MB"

[gaozig]

perfetto_trace_shell version
Perfetto v49.0-33a4fd078 (33a4fd0)
Trace Processor RPC API version: 14

[stevegolton]

It looks like the first part of 'Loading trace: XXX' message has been overwritten with 'Killed', which happens when the process is terminated. I'd suspect an OOM?

It looks like your trace at least 1248MB, which is on the large side, I'd expect you'll require a fair bit of available memory to be able to load it into trace processor.

[stevegolton]

I'll close this issue as it doesn't seem to be a problem with trace processor.

[gaozig]

the following version works pretty welll, no killed occur
Perfetto v39.0-d9ce4c0af (d9ce4c0)
Trace Processor RPC API version: 10

[gaozig]

@stevegolton

[stevegolton]

I wouldn't be surprised if more modern versions of TP use more memory. We typically optimize for speed, not memory footprint. It's also possible there could be a bug that just happens to be triggered by your trace, but I can't help you much given the information you've shared. If you want us to help you'll need to help us reproduce the bug by sending us:

• The offending trace.
• Your system specs.
• Kernel logs which could indicate whether an OOM has occurred and why.