Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[BUG] GPU buffer overrun in zstd::init_huff_tables #15096

Closed
jlowe opened this issue Feb 20, 2024 · 1 comment
Closed

[BUG] GPU buffer overrun in zstd::init_huff_tables #15096

jlowe opened this issue Feb 20, 2024 · 1 comment
Labels
bug Something isn't working cuIO cuIO issue libcudf Affects libcudf (C++/CUDA) code. Spark Functionality that helps Spark RAPIDS

Comments

@jlowe
Copy link
Member

jlowe commented Feb 20, 2024

Describe the bug
Encountered a GPU illegal address crash while trying to load a ZSTD-compressed Parquet file in Spark using the RAPIDS Accelerator. compute-sanitizer shows the crash originating in zstd::init_huff_tables.

Steps/Code to reproduce bug
The following C++ program will recreate the issue when run under compute-sanitizer with the attached Parquet data file as an argument (e.g..: compute-sanitizer ./repro pqdbg1637506717.parquet):

#include <cudf/io/parquet.hpp>
#include <iostream>

int main(int argc, char** argv) {
  if (argc != 2) {
    std::cerr << "Usage: " << argv[0] << " <parquet_file>" << std::endl;
    return 1;
  }
  cudf::io::parquet_reader_options read_opts =
    cudf::io::parquet_reader_options::builder(cudf::io::source_info{argv[1]});
  auto result = cudf::io::read_parquet(read_opts);
  return 0;
}

pqdbg1637506717.parquet.gz

Expected behavior
Parquet file should load without crashing.

@jlowe jlowe added bug Something isn't working libcudf Affects libcudf (C++/CUDA) code. cuIO cuIO issue Spark Functionality that helps Spark RAPIDS labels Feb 20, 2024
raydouglass pushed a commit to rapidsai/rapids-cmake that referenced this issue Feb 26, 2024
Move to nvcomp 3.0.6. Needed for a hotfix for rapidsai/cudf#15096.

Authors:
   - Bradley Dice (https://github.com/bdice)

Approvers:
   - Vyas Ramasubramani (https://github.com/vyasr)
raydouglass pushed a commit to rapidsai/kvikio that referenced this issue Feb 26, 2024
This PR bumps nvcomp to 3.0.6. This is needed as a hotfix for rapidsai/cudf#15096.

Depends on:
- conda-forge/nvcomp-feedstock#14
- rapidsai/rapids-cmake#542

Authors:
   - Bradley Dice (https://github.com/bdice)

Approvers:
   - Ray Douglass (https://github.com/raydouglass)
raydouglass pushed a commit that referenced this issue Feb 27, 2024
This PR bumps nvcomp to 3.0.6. This is needed as a hotfix for #15096.

Depends on:
- conda-forge/nvcomp-feedstock#14
- rapidsai/rapids-cmake#542
- rapidsai/kvikio#346

Authors:
   - Bradley Dice (https://github.com/bdice)

Approvers:
   - Lawrence Mitchell (https://github.com/wence-)
   - Ray Douglass (https://github.com/raydouglass)
@bdice
Copy link
Contributor

bdice commented Mar 10, 2024

Closed by #15128.

@bdice bdice closed this as completed Mar 10, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working cuIO cuIO issue libcudf Affects libcudf (C++/CUDA) code. Spark Functionality that helps Spark RAPIDS
Projects
None yet
Development

No branches or pull requests

2 participants