| Message ID | 20200902162805.200436-1-colin.king@canonical.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [next] RDMA/ucma: fix memory leak of mc on an xa_alloc failure | expand |
On Wed, Sep 02, 2020 at 05:28:05PM +0100, Colin King wrote: > From: Colin Ian King <colin.king@canonical.com> > > Currently when an xa_alloc failure occurs the error exit path leaks > the allocated object mc. Fix this by adding an error return path > that frees mc and rename error exit paths err3 to err4 and err2 to err3. > > Fixes: 95fe51096b7a ("RDMA/ucma: Remove mc_list and rely on xarray") > Signed-off-by: Colin Ian King <colin.king@canonical.com> > --- > drivers/infiniband/core/ucma.c | 11 ++++++----- > 1 file changed, 6 insertions(+), 5 deletions(-) Alex beat you too it, so I will use his patch, thanks Jason
diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c index f2c9ef6ae481..f081da35e3cf 100644 --- a/drivers/infiniband/core/ucma.c +++ b/drivers/infiniband/core/ucma.c @@ -1464,7 +1464,7 @@ static ssize_t ucma_process_join(struct ucma_file *file, if (xa_alloc(&multicast_table, &mc->id, NULL, xa_limit_32b, GFP_KERNEL)) { ret = -ENOMEM; - goto err1; + goto err2; } mutex_lock(&ctx->mutex); @@ -1472,13 +1472,13 @@ static ssize_t ucma_process_join(struct ucma_file *file, join_state, mc); mutex_unlock(&ctx->mutex); if (ret) - goto err2; + goto err3; resp.id = mc->id; if (copy_to_user(u64_to_user_ptr(cmd->response), &resp, sizeof(resp))) { ret = -EFAULT; - goto err3; + goto err4; } xa_store(&multicast_table, mc->id, mc, 0); @@ -1486,13 +1486,14 @@ static ssize_t ucma_process_join(struct ucma_file *file, ucma_put_ctx(ctx); return 0; -err3: +err4: mutex_lock(&ctx->mutex); rdma_leave_multicast(ctx->cm_id, (struct sockaddr *) &mc->addr); mutex_unlock(&ctx->mutex); ucma_cleanup_mc_events(mc); -err2: +err3: xa_erase(&multicast_table, mc->id); +err2: kfree(mc); err1: ucma_put_ctx(ctx);