similar to: [PATCH nbdkit] server: Remove tricksy initialization of struct b_conn_handle.

Also add an inline accessor function (get_handle). This also removes the unused function connection_get_handle. --- server/internal.h | 24 ++++++++++++++++----- server/backend.c | 50 ++++++++++++++++++++++---------------------- server/connections.c | 11 +--------- 3 files changed, 45 insertions(+), 40 deletions(-) diff --git a/server/internal.h b/server/internal.h index

We don't know how long a plugin's .get_size() will take, but we also documented that it shouldn't change per connection and therefore can be cached. It's not hard to see that we have to consult the size per connection (see commit b3a43ccd for a test that purposefully exposes different sizes to separate clients), nor to search the code to see we already cache it at the protocol

Similar to the previous patch in caching size, we want to avoid calling into the plugin more than once per connection on any of the flag determination callbacks. The following script demonstrates the speedup, where we avoid repeated calls into a slow can_fua. Pre-patch: $ cat script case "$1" in get_size) echo 1m;; can_fua) sleep 1; echo native;; can_write | can_zero |

It's confusing to use the same terminology for a single backend as for the linked list of backends. In particular it's often not clear if we're calling the next backend or the whole chain of backends. --- server/internal.h | 14 ++++++++++-- server/connections.c | 20 ++++++++--------- server/locks.c | 2 +- server/main.c

The retry filter was originally written to be the closest filter to the plugin, with no other filters in between; as such, the reopen command did not have to worry about recursion or about .prepare or .finalize. But it is not that much harder to properly track everything needed to allow other filters to be retried, as long as we are careful to never call .finalize unless .prepare succeeded, never

Since commit 86fdb48c6a5362d66865493d9d2172166f99722e we have stored the connection object in thread-local storage. In this very large, but mostly mechanical change we stop passing the connection pointer around everywhere, and instead use the value stored in thread-local storage. This assumes a 1-1 mapping between the connection and the current thread which is true in *most* places.

Low-level pthread locks should not fail except in extreme cases of programmer bugs; we're better off calling attention to such bugs rather than just assuming that they work and continuing on with possibly inconsistent state. Signed-off-by: Eric Blake <eblake@redhat.com> --- server/connections.c | 20 ++++++++++++-------- server/locks.c | 42

Previously only one handle could be stored, but we will need to store multiple handles when we have filters. The plugin handle is defined as index 0. Filters will use indices > 0. --- src/connections.c | 37 ++++++++++++++++++++++++++++++------- src/internal.h | 4 ++-- src/plugins.c | 53 +++++++++++++++++++++++++++-------------------------- 3 files changed, 59 insertions(+), 35

From: "Richard W.M. Jones" <rjones@redhat.com> Previously only one handle could be stored, but we will need to store multiple handles when we have filters. The plugin handle is defined as index 0. Filters will use indices > 0. Message-Id: <20180117205356.8699-6-rjones@redhat.com> [eblake: rework for FUA support] Signed-off-by: Eric Blake <eblake@redhat.com> ---

It's finally time to implement one of the TODO items: we want to support a thread pool of parallel readers from the client, in order to allow multiple in-flight operations with potential out-of-order completion. We also need at least one plugin that supports parallel processing for testing the option; the file plugin fits the bill. Add and document a new command line option, -t/--threads=N,

Updates commit 5cdf4d6c89bdb802be234f2fccc8157b7228b546 and commit a6b88b195a959b17524d1c8353fd425d4891dc5f. --- server/backend.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/server/backend.c b/server/backend.c index 64dbf7d..6c102f9 100644 --- a/server/backend.c +++ b/server/backend.c @@ -209,8 +209,7 @@ backend_close (struct backend *b, struct connection *conn) debug

Now instead of validating just the client's request, we are validating that all filters are passing valid requests on down. In protocol.c, we were able to assert that our computation of eflags populated all of the flags, and thus calls such as backend_can_write would not fail; however, with filters, keeping those assertions mean the burden is now on the filter to avoid calling into

For GCC only, define unlikely() macro. Use it on error paths to move code out of the hot path. In the server only, use the debug() macro (don't call nbdkit_debug directly). This macro checks the verbose flag and moves the call to nbdkit_debug out of the hot path. --- server/connections.c | 11 ++++++----- server/internal.h | 17 ++++++++++++++++- server/plugins.c | 2 +-

Mostly code motion. --- src/Makefile.am | 1 + src/connections.c | 14 +++---- src/internal.h | 14 ++++--- src/locks.c | 115 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ src/plugins.c | 77 +++++------------------------------- 5 files changed, 142 insertions(+), 79 deletions(-) diff --git a/src/Makefile.am b/src/Makefile.am index 12b9043..1f05eab 100644 ---

Mostly code motion. --- src/Makefile.am | 1 + src/connections.c | 14 +++---- src/internal.h | 14 ++++--- src/locks.c | 115 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ src/plugins.c | 77 +++++------------------------------- 5 files changed, 142 insertions(+), 79 deletions(-) diff --git a/src/Makefile.am b/src/Makefile.am index 12b9043..1f05eab 100644 ---

On 9/19/19 10:26 AM, Richard W.M. Jones wrote: > This is intended for use by the forthcoming retry filter to close and > reopen the backend chain. It is handled entirely by server/backend.c > as no cooperation is needed with the plugin. > > Note the explicit readonly parameter: An alternative would be to store > the previous readonly setting in the b_conn_handle struct.

This allows plugins (or filters) to read the export name which was passed to the server from the client. UNFINISHED: - Needs tests --- TODO | 8 ++++++ docs/nbdkit-plugin.pod | 39 ++++++++++++++++++++++++++++ server/connections.c | 1 + server/internal.h | 1 + server/protocol-handshake-newstyle.c | 30

Most known NBD clients do not bother with NBD_OPT_INFO (except for clients like 'qemu-nbd --list' that don't ever intend to connect), but go straight to NBD_OPT_GO. However, it's not too hard to hack up qemu to add in an extra client step (whether info on the same name, or more interestingly, info on a different name), as a patch against qemu commit 6f214b30445: | diff --git

This is intended for use by the forthcoming retry filter to close and reopen the backend chain. It is handled entirely by server/backend.c as no cooperation is needed with the plugin. Note the explicit readonly parameter: An alternative would be to store the previous readonly setting in the b_conn_handle struct. However passing it explicitly allows the retry filter to retry as readonly, which

This returns the plugin name, which for plugins is the same as the ordinary .name method (but for filters will be different). --- src/connections.c | 4 ++-- src/internal.h | 1 + src/plugins.c | 1 + 3 files changed, 4 insertions(+), 2 deletions(-) diff --git a/src/connections.c b/src/connections.c index 921a5b2..e1ac543 100644 --- a/src/connections.c +++ b/src/connections.c @@ -221,7