similar to: [nbdkit PATCH 0/2] more fast zero prep

After more than a month since v2 [1], I've finally got my FUA support series polished. This is all of my outstanding patches, even though some of them were originally posted in separate threads from the original FUA post [2], [3] [1] https://www.redhat.com/archives/libguestfs/2018-January/msg00113.html [2] https://www.redhat.com/archives/libguestfs/2018-January/msg00219.html [3]

It's easy to use the sh script to demonstrate that nbdkit is inefficiently calling into .get_size, .can_fua, and friends more than necessary. We've also commented on the list in the past that it would be nice to ensure that when filters call into next_ops, they are not violating constraints (as we've have to fix several bugs in the past where we did not have such checking to protect

This is a cover letter to a series of patches being proposed in tandem to four different projects: - nbd: Document a new NBD_CMD_FLAG_FAST_ZERO command flag - qemu: Implement the flag for both clients and server - libnbd: Implement the flag for clients - nbdkit: Implement the flag for servers, including the nbd passthrough client If you want to test the patches together, I've pushed a

I still need to add testsuite coverage. Perhaps it might be easier if I create a new '--filter=log logfile=foo' filter that produces a log of which commands a client sent, then compare the log using a known client that uses write_zeroes (qemu-io works well) both with and without --filter=nozero to prove that the change in advertisement changes the commands sent over the wire (that would

It may be useful to test whether the client's use of NBD_CMD_FLAG_NO_HOLE makes a difference; do this by adding a mode to --filter=nozero to force a non-trimming zero write. Signed-off-by: Eric Blake <eblake@redhat.com> --- filters/nozero/nbdkit-nozero-filter.pod | 19 +++++++---- filters/nozero/nozero.c | 45 ++++++++++++++++++++----- tests/test-nozero.sh

I still don't have .prepare/.finalize working cleanly across reopen, but did find a nasty bug where a botched assertion means we failed to notice reads beyond EOF in both the xz and retry filter. Refactoring backend.c will make .finalize work easier. Eric Blake (5): xz: Avoid reading beyond EOF retry: Check size before transactions tests: Test retry when get_size values change

v1 was here: https://www.redhat.com/archives/libguestfs/2020-February/msg00081.html v2 replaces struct connection *conn = GET_CONN; with GET_CONN; which sets conn implicitly and asserts that it is non-NULL. If we actually want to test if conn is non-NULL or behave differently, then you must use threadlocal_get_conn() instead, and some existing uses do that. Rich.

The third patch is a large but mechanical change which gets rid of passing around struct connection * entirely within the server, preferring instead to reference the connection through thread-local storage. I hope this is a gateway to simplifying other parts of the code. Rich.

This is a subset of the last half of the larger 9-patch series. The uncontroversial first half of that series is pushed, but here, I tried to reduce the size of the patches by splitting out some of the more complex changes, so that the rest of the changes remaining in the series are more mechanical. In turn, it forced me to write timing tests, which let me spot another spot where we are wasting

The nozero filter can split a large request with FUA into several smaller requests; optimize whether the FUA flag is passed on to the next layer based on whether FUA is emulated with flush (only the last write needs it) or is natively supported (every write needs it). Missed in commit df0cc21d. Signed-off-by: Eric Blake <eblake@redhat.com> --- filters/nozero/nozero.c | 17 +++++++++++++++--

Hi all, It seems GCC 9 does not put read-only zero-initialized variables into the BSS section anymore; instead it is put into RODATA. (See the thread at [0], especially [1]) In filter/nozero a 64M large, static, zeroed, read-only array is declared. The new behavior of GCC puts this array as-is into the binary inflating the size by a factor of around 10000. (Clang 8 and older GCCs work

Two hopefully useful additions to the fua filter. The second one is kind of like cache=unsafe in qemu, in that it exchanges correctness for speed. Useful for data which is easily recreated in the event of a crash or for people who like living on the edge and have good backups. Rich.

v4: https://www.redhat.com/archives/libguestfs/2019-January/msg00032.html v5: - Now we set the block size at run time. I'd like to say that I was able to test this change, but unfortunately I couldn't find any easy way to create a filesystem on x86-64 with a block size > 4K. Ext4 doesn't support it at all, and XFS doesn't support block size > page size (and I

Since v1: - rework .can_cache to be tri-state, with default of no advertisement (ripple effect through other patches) - add a lot more patches in order to round out filter support And in the meantime, Rich pushed NBD_CMD_CACHE support into libnbd, so in theory we now have a way to test cache commands through the entire stack. Eric Blake (24): server: Internal hooks for implementing

Based on Rich's review of my v1 that touched only cache.c, I have now tried to bring all three filters with alignment rounding in line with one another. There is definitely room for future improvements once we teach nbdkit to let filters and plugins advertise block sizes, but I'm hoping to get NBD_CMD_CACHE implemented first. Eric Blake (2): blocksize: Process requests in linear order

Rebased filters patch. Requires current git master + the locks / thread model fix (https://www.redhat.com/archives/libguestfs/2018-January/msg00128.html) So a few changes here since last time: The "introduce filters" and "implement filters" patches are squashed together. I introduced a concept of .prepare and .finalize. These run before and after the data serving phase

Here's what I'm currently playing with; I'm not ready to commit anything until I rebase my FUA work on top of this, as I only want to break filter ABI once between releases. Eric Blake (2): backend: Rework internal/filter error return semantics filters: Add log filter TODO | 2 - docs/nbdkit-filter.pod | 84 +++++++-- docs/nbdkit.pod

First thing to say is that I need to do a *lot* more testing on this, so this is just an early peek. In particular, although it passed ‘make check && make check-valgrind’ I have *not* tested it against a multi-conn-aware client such as the Linux kernel >= 4.9. This implements NBD_FLAG_CAN_MULTI_CONN, described in the protocol doc as: "NBD_FLAG_CAN_MULTI_CONN: Indicates that

For existing commits, this is almost identical to v1, except that I updated some commit messages and reordered the commits in a somewhat more logical sequence. The main changes are the extra commits: [06/11] plugins: Return NBD_FLAG_CAN_MULTI_CONN from some readonly plugins. - Readonly plugins that can set the flag unconditionally. [09/11] partitioning: Return NBD_FLAG_CAN_MULTI_CONN. [10/11]

I'm still working on another set of patches to have reopen call .finalize/.prepare (so that another filter can safely appear between retry and the plugin), but for tonight, these are the patches I think are ready to go. Eric Blake (4): retry: Handle can_fua and can_fast_zero changes tests: Test retry with different fua/fast-zero flags server: Close backends if a filter's .open fails