I'm not sure if it is a PDFium issue actually.

yangguo@, could you please take a look if it is an internal V8 error or incorrect usage from PDFium side?

why would you run a libfuzzer test with lsan? it's not supposed to free up stuff ever.

libfuzzer tests have an initialize hook, but no deinitialize hook - they're designed to leak memory.

jochen, no, fuzz tartgets are not expected to leak memory in the lsan sense. 
The one-time initialization performed at process startup is not a leak as long as 
the pointers to the allocated objects are not lost. 

mmoroz@, how do I reproduce this? 
I've tried this: 
% git pull && gclient sync
% gn gen out/libfuzzer '--args=use_libfuzzer=true is_asan=true enable_nacl=false' --check
% ninja -C out/libfuzzer pdfium_fuzzer
% ASAN_OPTIONS=detect_leaks=1:symbolize=1  ./out/libfuzzer/pdfium_fuzzer ~/Downloads/260560b25029714bea2854172dc4dbee2b3b038c 
INFO: Seed: 3926530079
./out/libfuzzer/pdfium_fuzzer: Running 1 inputs 1 time(s) each.
/usr/local/google/home/kcc/Downloads/260560b25029714bea2854172dc4dbee2b3b038c: 7 ms

(no leak report). 

When doing the actual fuzzing I do see some leak reports elsewhere (e.g. CPDF_Parser::StartParse), but not in v8. 

Good question, I cannot reproduce it too...

Let's keep it closed then. 
I don't think it was a false positive though -- probably just got fixed. 

ClusterFuzz has detected this issue as fixed in range 395675:395769.

Detailed report: https://cluster-fuzz.appspot.com/testcase?key=5107287554785280

Fuzzer: pdfium_fuzzer
Job Type: libfuzzer_chrome_asan
Platform Id: linux

Crash Type: Direct-leak
Crash Address: 
Crash State:
  Update
  v8::internal::Deserializer::Allocate
  v8::internal::Deserializer::ReadObject
  
Regressed: https://cluster-fuzz.appspot.com/revisions?job=libfuzzer_chrome_asan&range=375690:375725
Fixed: https://cluster-fuzz.appspot.com/revisions?job=libfuzzer_chrome_asan&range=395675:395769

Unminimized Testcase: https://cluster-fuzz.appspot.com/download/AMIfv96KhTrHHZjtYPEGay18iXu5uAnwmagFC1yPCicZRjhPUJyRIcUG5GDMFL4o-sEL6kzbYTxt9SdiXY8Op49dsZGk3BizTGijFgvAZIuAPvnHMr5QQctd141MzgIY4B4Goo2oPRqXYZRVt71XYhESHMW0VZ2I4Q?testcase_id=5107287554785280


See https://chromium.googlesource.com/chromium/src/+/master/testing/libfuzzer/reproducing.md for more information.

If you suspect that the result above is incorrect, try re-doing that job on the test case report page.

Removing EditIssue view restrictions from ClusterFuzz filed bugs. If you believe that this issue should still be restricted, please reapply the label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot