Even though v5 was posted earlier today, it was worth a respin: - 2/6: add R-b [Berto] - 4/6, 6/6: improve commit messages [Max] - 5/6: new patch, with an iotests proving that 4/6 is a bug fix [Max]
The new test is rather slow (nearly 90 seconds for me using tmpfs) unless it skips entirely (such as testing on ext4); ideas for speeding it up are welcome (translation: maybe qemu should optimize the search for the next available cluster to allocate, and/or qemu-img check should be faster at reporting leaked clusters) 001/6:[----] [--] 'qcow2: Prefer byte-based calls into bs->file' 002/6:[----] [--] 'qcow2: Document some maximum size constraints' 003/6:[----] [--] 'qcow2: Reduce REFT_OFFSET_MASK' 004/6:[----] [--] 'qcow2: Don't allow overflow during cluster allocation' 005/6:[down] 'iotests: Add new test 214 for max compressed cluster offset' 006/6:[----] [--] 'qcow2: Avoid memory over-allocation on compressed images' Eric Blake (6): qcow2: Prefer byte-based calls into bs->file qcow2: Document some maximum size constraints qcow2: Reduce REFT_OFFSET_MASK qcow2: Don't allow overflow during cluster allocation iotests: Add new test 214 for max compressed cluster offset qcow2: Avoid memory over-allocation on compressed images docs/interop/qcow2.txt | 40 +++++++++++++++++-- block/qcow2.h | 8 +++- block/qcow2-cluster.c | 32 +++++++++------ block/qcow2-refcount.c | 27 ++++++++----- block/qcow2.c | 2 +- tests/qemu-iotests/214 | 97 ++++++++++++++++++++++++++++++++++++++++++++++ tests/qemu-iotests/214.out | 54 ++++++++++++++++++++++++++ tests/qemu-iotests/group | 1 + 8 files changed, 234 insertions(+), 27 deletions(-) create mode 100755 tests/qemu-iotests/214 create mode 100644 tests/qemu-iotests/214.out -- 2.14.3