diff options
author | Yordan Karadzhov (VMware) <y.karadz@gmail.com> | 2021-05-17 17:21:37 +0300 |
---|---|---|
committer | Yordan Karadzhov (VMware) <y.karadz@gmail.com> | 2021-05-18 10:52:15 +0300 |
commit | d0c2193ef1d6d7cfe613b3c54ac732ebe76d60e8 (patch) | |
tree | 0f4504a8da6161cd42eeee0cf7009976c4230b22 | |
parent | 0929c4b624f83a3f663cf31b09769ddfce0bc91a (diff) | |
download | kernel-shark-d0c2193ef1d6d7cfe613b3c54ac732ebe76d60e8.tar.gz |
kernel-shark: Better handling of plugins when appending data file
It may sound like a good idea to reinitialize all plugins for all
existing Data streams after a new stream is appended, but it isn't.
Such reset will re-initialize all Data containers open by the
plugins, hence it requires to do a complete reload of all previously
loaded data. Currently we reset the plugins but do not reload the
data, which is a bug. Because, reloading the data can be slow on
large data sets, I prefer to fix the bug, by eliminating the reload
of the plugins.
Link: https://lore.kernel.org/linux-trace-devel/20210517142140.286153-5-y.karadz@gmail.com
Signed-off-by: Yordan Karadzhov (VMware) <y.karadz@gmail.com>
-rw-r--r-- | src/KsUtils.cpp | 3 |
1 files changed, 1 insertions, 2 deletions
diff --git a/src/KsUtils.cpp b/src/KsUtils.cpp index 757f49c3..3db89510 100644 --- a/src/KsUtils.cpp +++ b/src/KsUtils.cpp @@ -677,8 +677,7 @@ int KsDataStore::_openDataFile(kshark_context *kshark_ctx, if (kshark_is_tep(kshark_ctx->stream[sd])) { kshark_tep_init_all_buffers(kshark_ctx, sd); - for (int i = 0; i < kshark_ctx->n_streams; ++i) - kshark_tep_handle_plugins(kshark_ctx, i); + kshark_tep_handle_plugins(kshark_ctx, sd); } return sd; |