diff options
author | Jason Gunthorpe <jgg@mellanox.com> | 2019-08-21 20:12:29 +0300 |
---|---|---|
committer | Jason Gunthorpe <jgg@mellanox.com> | 2019-08-22 02:58:18 +0300 |
commit | daa138a58c802e7b4c2fb73f9b85bb082616ef43 (patch) | |
tree | be913e8e3745bb367d2ba371598f447649102cfc /tools/perf/util/thread.c | |
parent | 6869b7b206595ae0e326f59719090351eb8f4f5d (diff) | |
parent | fba0e448a2c5b297a4ddc1ec4e48f4aa6600a1c9 (diff) | |
download | linux-daa138a58c802e7b4c2fb73f9b85bb082616ef43.tar.xz |
Merge branch 'odp_fixes' into hmm.git
From rdma.git
Jason Gunthorpe says:
====================
This is a collection of general cleanups for ODP to clarify some of the
flows around umem creation and use of the interval tree.
====================
The branch is based on v5.3-rc5 due to dependencies, and is being taken
into hmm.git due to dependencies in the next patches.
* odp_fixes:
RDMA/mlx5: Use odp instead of mr->umem in pagefault_mr
RDMA/mlx5: Use ib_umem_start instead of umem.address
RDMA/core: Make invalidate_range a device operation
RDMA/odp: Use kvcalloc for the dma_list and page_list
RDMA/odp: Check for overflow when computing the umem_odp end
RDMA/odp: Provide ib_umem_odp_release() to undo the allocs
RDMA/odp: Split creating a umem_odp from ib_umem_get
RDMA/odp: Make the three ways to create a umem_odp clear
RMDA/odp: Consolidate umem_odp initialization
RDMA/odp: Make it clearer when a umem is an implicit ODP umem
RDMA/odp: Iterate over the whole rbtree directly
RDMA/odp: Use the common interval tree library instead of generic
RDMA/mlx5: Fix MR npages calculation for IB_ACCESS_HUGETLB
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Diffstat (limited to 'tools/perf/util/thread.c')
-rw-r--r-- | tools/perf/util/thread.c | 12 |
1 files changed, 11 insertions, 1 deletions
diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c index 873ab505ca80..590793cc5142 100644 --- a/tools/perf/util/thread.c +++ b/tools/perf/util/thread.c @@ -214,14 +214,24 @@ struct comm *thread__comm(const struct thread *thread) struct comm *thread__exec_comm(const struct thread *thread) { - struct comm *comm, *last = NULL; + struct comm *comm, *last = NULL, *second_last = NULL; list_for_each_entry(comm, &thread->comm_list, list) { if (comm->exec) return comm; + second_last = last; last = comm; } + /* + * 'last' with no start time might be the parent's comm of a synthesized + * thread (created by processing a synthesized fork event). For a main + * thread, that is very probably wrong. Prefer a later comm to avoid + * that case. + */ + if (second_last && !last->start && thread->pid_ == thread->tid) + return second_last; + return last; } |