Rte_memcpy need not be used for the simple case of copying a fix sized because it is not faster and will hide problems from code analysis tools. Static and dynamic analyzers special case memcpy() and can find issues that rte_memcpy() will hide.
Note: many platforms use memcpy instead of rte_memcpy already. The only reason rte_memcpy() was added in original DPDK was that the older Gcc was doing slower on x86 than the special rte_memcpy() when handling larger size copies. Signed-off-by: Stephen Hemminger <[email protected]> --- devtools/cocci/rte_memcpy_fixed.cocci | 15 +++++++++++++++ 1 file changed, 15 insertions(+) create mode 100644 devtools/cocci/rte_memcpy_fixed.cocci diff --git a/devtools/cocci/rte_memcpy_fixed.cocci b/devtools/cocci/rte_memcpy_fixed.cocci new file mode 100644 index 0000000000..1f264711eb --- /dev/null +++ b/devtools/cocci/rte_memcpy_fixed.cocci @@ -0,0 +1,15 @@ +// +// rte_memcpy should not be used for simple fixed size structure +// because compiler's are smart enough to inline these. +// +@@ +expression src, dst, E; +constant size; +@@ +( +- rte_memcpy(dst, src, sizeof(E)) ++ memcpy(dst, src, sizeof(E)) +| +- rte_memcpy(dst, src, size) ++ memcpy(dst, src, size) +) -- 2.53.0
