search for: 45362d4

...mds_in_flight list does not have to care about maintaining FIFO ordering. --- generator/states-reply.c | 13 ++++++++++--- lib/rw.c | 13 ++++++++++--- 2 files changed, 20 insertions(+), 6 deletions(-) diff --git a/generator/states-reply.c b/generator/states-reply.c index 93f6cda..45362d4 100644 --- a/generator/states-reply.c +++ b/generator/states-reply.c @@ -103,13 +103,20 @@ } assert (cmd != NULL); - /* Move it to the cmds_done list. */ + /* Move it to the end of the cmds_done list. */ if (prev_cmd != NULL) prev_cmd->next = cmd->next; else conn->...

...+C<aio_command_completed> to actually retire the command and learn +whether the command was successful."; + }; + "connection_state", { default_call with args = []; ret = RConstString; diff --git a/generator/states-reply.c b/generator/states-reply.c index 93f6cda..45362d4 100644 --- a/generator/states-reply.c +++ b/generator/states-reply.c @@ -103,13 +103,20 @@ } assert (cmd != NULL); - /* Move it to the cmds_done list. */ + /* Move it to the end of the cmds_done list. */ if (prev_cmd != NULL) prev_cmd->next = cmd->next; else conn->...

...START); + return 0; + ISSUE_COMMAND.FINISH: struct command_in_flight *cmd; + assert (!conn->wlen); assert (conn->cmds_to_issue != NULL); cmd = conn->cmds_to_issue; conn->cmds_to_issue = cmd->next; diff --git a/generator/states-reply.c b/generator/states-reply.c index 45362d4..6bb503a 100644 --- a/generator/states-reply.c +++ b/generator/states-reply.c @@ -118,7 +118,10 @@ else conn->cmds_done = cmd; - SET_NEXT_STATE (%.READY); + if (conn->cmds_to_issue) + SET_NEXT_STATE (%^ISSUE_COMMAND.START); + else + SET_NEXT_STATE (%.READY); return 0; }...

This might not be the final solution, but it certainly seems to solve a deadlock for me that I could trigger by using 'nbdkit --filter=noparallel memory 512k' and calling nbd_aio_pread for a request larger than 256k (enough for the Linux kernel to block the server until libnbd read()s), immediately followed by nbd_aio_pwrite for a request larger than 256k (enough to block libnbd until the

Since v2: - rebase to Rich's new API calls - more refactoring in patch 1 (retitled) - new patches 3 and 4 - fix data corruption in patch 6 (was 4) - more tweaks to the reproducer example (including using new API from 3) Eric Blake (7): lib: Refactor command_common() to do more common work commands: Allow for a command queue commands: Expose FIFO ordering of server completions

This isn't quite finished because not all of the tests or examples have been updated, but it demonstrates an idea: Should we forget about the concept of having multiple connections managed under a single handle? In this patch there is a single ‘struct nbd_handle *’ which manages a single state machine and connection (and therefore no nbd_connection). To connect to a multi-conn server you must