similar to: [PATCH nbdkit 3/4] python: Allow thread model to be set from Python plugins.

This picks up python and ocaml. Some of our languages are lacking a number of bindings (for example, lua and perl lack .extents, so I didn't have anything to copy from), and I felt less comfortable with golang and rust. But for python and ocaml, I was able to test a working implementation. Eric Blake (2): python: Implement .list_exports and friends ocaml: Implement .list_exports and

Patch 2 certainly allows you to set the thread model. However patch 3 shows that if you set it to nbdkit.THREAD_MODEL_PARALLEL it will crash. If you look closely at the stack trace (attached below) you can see that ignoring threads which are in parts of nbdkit unrelated to Python: Thread 4: In pread, waiting in time.sleep(). This thread has released the GIL. Thread 2: Started to

When I added .can_FOO caching in 1.16, I missed the case that the sh plugin itself was calling .can_flush twice in some situations (in order to default .can_fua). Then right after, I regressed it to call .can_zero twice (in order to default .can_fast_zero). I also missed that .thread_model could use better caching, because at the time, I did not add testsuite coverage. Fix that now. Eric Blake

I like this change. I think we were overloading the config_complete method before to do two different things (complete configuration; do any allocation/housekeeping necessary before we can start serving). The only questions in my mind are whether we want this before 1.18, and whether the name ("get_ready") is a good one. Rich.

And fill out most of the missing bits of the API. Rich.

--- plugins/python/nbdkit-python-plugin.pod | 19 ++++++- plugins/python/python.c | 75 +++++++++++++++++++++++++ tests/test-python-plugin.py | 8 ++- tests/test_python.py | 73 +++++++++++++++++++++++- 4 files changed, 169 insertions(+), 6 deletions(-) diff --git a/plugins/python/nbdkit-python-plugin.pod b/plugins/python/nbdkit-python-plugin.pod

v1: https://www.redhat.com/archives/libguestfs/2019-November/msg00153.html v2: - Fix implementation of can_cache. - Add implementation of can_fua. - Add a very thorough test suite which tests every command + flag combination.

As mentioned in my reviews, I wonder if we should make our python callbacks look a bit more Pythonic by having kwargs added for each new flag that we want to expose. The idea was first floated here: https://www.redhat.com/archives/libguestfs/2018-April/msg00108.html Note that with my proposal, there is no need for a python script to expose a global API_VERSION variable; new flags are added

On 3/18/20 8:21 PM, Eric Blake wrote: > In commit c306fa93ab and neighbors (v1.15.1), a concerted effort went > into caching the results of .can_FOO callbacks, with commit messages > demonstrating that a plugin with a slow callback should not have that > delay magnified multiple times. But nothing was added to the > testsuite at the time, and with the sh and eval plugins, we still

Fairly straightforward. .list_exports uses the same idiom as .extents for returning an iterable of tuples, with additional support for a bare name rather than a name/desc tuple. .default_export and .export_description are rather easy clients of nbdkit_strdup_intern. Signed-off-by: Eric Blake <eblake at redhat.com> --- plugins/python/nbdkit-python-plugin.pod | 25 ++++ tests/Makefile.am

v3 was here: https://www.redhat.com/archives/libguestfs/2019-November/msg00209.html In v4: - Rebase on top of current master. Includes various fixes and updates required because of Nir's patches that went into master. - Fix api_version() -> API_VERSION in patch 2 noted previously on the mailing list. Rich.

v2 was here: https://www.redhat.com/archives/libguestfs/2019-November/msg00163.html I pushed patch 1 (with spelling fix), patch 4 and patch 5 since those were previously ACKed on the list. Differences in v3: - Add error checking to PyModule_AddIntConstant. - Use API_VERSION constant instead of function. - Add max API version supported to --dump-plugin output. - Print API_VERSION selected by

First out of our four language bindings to add FUA support (for reference, I added 'zero' support for python, perl, and ruby back in 1.1.13, then Rich had to add it for ocaml in 1.1.20). I tested this heavily under python 2, but for now only compile tested under python 3; I plan to do further testing there and make any tweaks if necessary. I wrote patch 5 early on, but then realized I

It's been several weeks since I posted v2 (I got distracted by improving libnbd to better test things, which in turn surfaced some major memory leak problems in nbdsh that are now fixed). Many of the patches are minor rebases from v2, with the biggest changes being fallout from: - patch 2: rename nbdkit_add_default_export to nbdkit_use_default_export - overall: this missed 1.22, so update

This is a major rewrite compared to my v1 series, where I've tried a lot harder to ensure that we still accommodate building on Haiku (although I have not actually yet fired up a Haiku VM to try it for myself). I also managed to make the sh plugin fully parallel, on capable platforms. See also my question on patch 10 on whether I've picked the best naming convention. Eric Blake (17):

Rich pointed out that making thread_model dynamic even for plugins makes some sense, so here is the code for doing it for 'sh'. I'm less confident on how to do it for OCaml and Rust (not to mention that those allow the plugin to already compile in their model, rather than the language binding glue forcing a model). The other languages (lua, perl, python, ruby) still need to be

I noticed these while working on adding fua support into python, these are independent enough to push now (and I'll have to rebase my 'optional may_trim' patch on top of this). Eric Blake (2): python: Use Py_XDEFREF() python: Simplify calling into plugin plugins/python/python.c | 106 ++++++++---------------------------------------- 1 file changed, 18 insertions(+), 88

The messages are not really useful to us, but they do bloat the debugging output of virt-v2v massively: nbdkit: python[1]: debug: acquire global request lock nbdkit: python[1]: debug: acquire per-connection request lock nbdkit: python[1]: debug: acquire unload prevention lock nbdkit: python[1]: debug: pwrite count=2097152 offset=4628414464 fua=0 nbdkit: python[1]: debug: release unload prevention

On platforms that lack atomic CLOEXEC, it's simpler to just always force serialization (no client thread will be executing when nbdkit itself is creating a new file descriptor) than to audit which plugins actually care about not getting any leaked fds. We have one final function that we need to use for atomic CLOEXEC; the next patch will actually put accept4() to use. Maybe this penalization

On Fri, Aug 02, 2019 at 02:26:11PM -0500, Eric Blake wrote: > Allow a plugin field to declare whether a parallel plugin can tolerate > windows where fds are not CLOEXEC, or must take precautions to avoid > leaking fds if the plugin may fork. For safety reasons, the flag > defaults to off, but many in-tree plugins can set it to on (most > commonly because they don't fork after