[Mesa-dev] [PATCH] etnaviv: Use reentrant screen lock around flush

Marek Vasut marex at denx.de
Mon Jun 3 23:40:51 UTC 2019

The flush callback may be called on the same pipe context, and thus
the same stream, from two different threads of execution. However,
etna_cmd_stream_flush{,2}() must not be called on the same stream
from two different threads of execution as that would mess up the
etna_bo refcounting and likely have other ugly side effects.

Fix this by using a reentrant screen lock around the flush callback.

Signed-off-by: Marek Vasut <marex at denx.de>
Cc: Christian Gmeiner <christian.gmeiner at gmail.com>
Cc: Lucas Stach <l.stach at pengutronix.de>
 src/gallium/drivers/etnaviv/etnaviv_context.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/src/gallium/drivers/etnaviv/etnaviv_context.c b/src/gallium/drivers/etnaviv/etnaviv_context.c
index a59338490b6..ab6b2d5e154 100644
--- a/src/gallium/drivers/etnaviv/etnaviv_context.c
+++ b/src/gallium/drivers/etnaviv/etnaviv_context.c
@@ -310,8 +310,11 @@ etna_flush(struct pipe_context *pctx, struct pipe_fence_handle **fence,
            enum pipe_flush_flags flags)
    struct etna_context *ctx = etna_context(pctx);
+   struct etna_screen *screen = ctx->screen;
    int out_fence_fd = -1;
+   mtx_lock(&screen->lock);
    list_for_each_entry(struct etna_hw_query, hq, &ctx->active_hw_queries, node)
       etna_hw_query_suspend(hq, ctx);
@@ -324,6 +327,8 @@ etna_flush(struct pipe_context *pctx, struct pipe_fence_handle **fence,
    if (fence)
       *fence = etna_fence_create(pctx, out_fence_fd);
+   mtx_unlock(&screen->lock);
 static void

More information about the mesa-dev mailing list