On 10-Jul-20 5:17 PM, Thomas Monjalon wrote:
10/07/2020 17:39, Burakov, Anatoly:
On 10-Jul-20 12:53 PM, Thomas Monjalon wrote:
The function pci_map_resource() returns MAP_FAILED in case of error.
When replacing the call to mmap() by rte_mem_map(),
the error code became NULL, breaking the API.
This function is probably not used outside of DPDK,
but it is still a problem for two reasons:
- the deprecation process was not followed
- the Linux function pci_vfio_mmap_bar() is broken for i40e
The error code is reverted to the Unix value MAP_FAILED.
Windows needs to define this special value (-1 as in Unix).
After proper deprecation process, the API could be changed again
if really needed.
Because of the switch from mmap() to rte_mem_map(),
another part of the API was changed: "int additional_flags"
are defined as "additional flags for the mapping range"
without mentioning it was directly used in mmap().
Currently it is directly used in rte_mem_map(),
that's why the values rte_map_flags must be mapped (sic) on the mmap ones
in case of Unix OS.
These are side effects of a badly defined API using Unix values.
[...]
/** Additional flags for memory mapping. */
enum rte_map_flags {
+#ifdef RTE_EXEC_ENV_WINDOWS
/** Changes to the mapped memory are visible to other processes. */
RTE_MAP_SHARED = 1 << 0,
/** Mapping is not backed by a regular file. */
@@ -35,6 +37,12 @@ enum rte_map_flags {
* it is not required to do so, thus mapping with this flag may fail.
*/
RTE_MAP_FORCE_ADDRESS = 1 << 3
+#else /* map mmap flags because they are exposed in pci_map_resource() API */
+ RTE_MAP_SHARED = MAP_SHARED,
+ RTE_MAP_ANONYMOUS = MAP_ANONYMOUS,
+ RTE_MAP_PRIVATE = MAP_PRIVATE,
+ RTE_MAP_FORCE_ADDRESS = MAP_FIXED,
+#endif
I'm probably missing something, but why is this needed? Doesn't
Yes you missed reading the commit log :)
Or maybe it is not written clearly enough. Will try to rephrase.
rte_mem_map() automatically translate these flags into proper ones?
pci_map_resource() will call rte_mem_map(), and that will translate
these flags into their Unix equivalents.
The problem is that we have an API which is taking mmap flags as input.
"int additional_flags" is a parameter of the function,
and are supposed to be mmap flags. But it is not stated clearly.
When Windows will use this function, it won't use mmap flags
but RTE_MAP_*. So we must accept both.
That's why the best is to make values the same.
In 20.11, we could change the API,
make clear that only RTE_MAP_* is accepted,
and remove this workaround.
Or even better, remove pci_map_resource from the PCI lib,
and implement it in the PCI bus driver.
pci_map_resource() function is a bad designed API
Right, this makes it clearer :) Thanks!
--
Thanks,
Anatoly
