eal/windows: fix IOVA mode detection and handling

Windows EAL did not detect IOVA mode and worked incorrectly
if physical addresses could not be obtained
(if virt2phys driver was missing or inaccessible).
In this case, rte_mem_virt2iova() reported RTE_BAD_IOVA for any address.
Inability to obtain IOVA, be it PA or VA, should cause a failure
for the DPDK allocator, but it was hidden by the implementation,
so allocations did not fail when they should.
The mode when DPDK cannot obtain PA but can work is IOVA-as-VA mode.
However, rte_eal_iova_mode() always returned RTE_IOVA_DC
(while it should only ever return RTE_IOVA_PA or RTE_IOVA_VA),
because IOVA mode detection was not implemented.

Implement IOVA mode detection:
1. Always allow to force --iova-mode=va.
2. Allow to force --iova-mode=pa only if virt2phys is available.
3. If no mode is forced and virt2phys is available,
   select the mode according to bus requests, default to PA.
4. If no mode is forced but virt2phys is unavailable, default to VA.
Fix rte_mem_virt2iova() by returning VA when using IOVA-as-VA.
Fix rte_eal_iova_mode() by returning the selected mode.

Fixes: 2a5d547a4a9b ("eal/windows: implement basic memory management")
Cc: stable@dpdk.org
Reported-by: Tal Shnaiderman <talshn@nvidia.com>
Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
Tested-by: Pallavi Kadam <pallavi.kadam@intel.com>
Acked-by: Pallavi Kadam <pallavi.kadam@intel.com>

diff --git a/lib/eal/windows/eal.c b/lib/eal/windows/eal.c
index 3d8c520..f7ce1b6 100644 (file)
--- a/lib/eal/windows/eal.c
+++ b/lib/eal/windows/eal.c
@@ -276,6 +276,8 @@ rte_eal_init(int argc, char **argv)
        const struct rte_config *config = rte_eal_get_configuration();
        struct internal_config *internal_conf =
                eal_get_internal_configuration();
+       bool has_phys_addr;
+       enum rte_iova_mode iova_mode;
        int ret;
 
        eal_log_init(NULL, 0);
@@ -322,18 +324,59 @@ rte_eal_init(int argc, char **argv)
                        internal_conf->memory = MEMSIZE_IF_NO_HUGE_PAGE;
        }
 
+       if (rte_eal_intr_init() < 0) {
+               rte_eal_init_alert("Cannot init interrupt-handling thread");
+               return -1;
+       }
+
+       if (rte_eal_timer_init() < 0) {
+               rte_eal_init_alert("Cannot init TSC timer");
+               rte_errno = EFAULT;
+               return -1;
+       }
+
+       bscan = rte_bus_scan();
+       if (bscan < 0) {
+               rte_eal_init_alert("Cannot scan the buses");
+               rte_errno = ENODEV;
+               return -1;
+       }
+
        if (eal_mem_win32api_init() < 0) {
                rte_eal_init_alert("Cannot access Win32 memory management");
                rte_errno = ENOTSUP;
                return -1;
        }
 
+       has_phys_addr = true;
        if (eal_mem_virt2iova_init() < 0) {
                /* Non-fatal error if physical addresses are not required. */
-               RTE_LOG(WARNING, EAL, "Cannot access virt2phys driver, "
+               RTE_LOG(DEBUG, EAL, "Cannot access virt2phys driver, "
                        "PA will not be available\n");
+               has_phys_addr = false;
        }
 
+       iova_mode = internal_conf->iova_mode;
+       if (iova_mode == RTE_IOVA_PA && !has_phys_addr) {
+               rte_eal_init_alert("Cannot use IOVA as 'PA' since physical addresses are not available");
+               rte_errno = EINVAL;
+               return -1;
+       }
+       if (iova_mode == RTE_IOVA_DC) {
+               RTE_LOG(DEBUG, EAL, "Specific IOVA mode is not requested, autodetecting\n");
+               if (has_phys_addr) {
+                       RTE_LOG(DEBUG, EAL, "Selecting IOVA mode according to bus requests\n");
+                       iova_mode = rte_bus_get_iommu_class();
+                       if (iova_mode == RTE_IOVA_DC)
+                               iova_mode = RTE_IOVA_PA;
+               } else {
+                       iova_mode = RTE_IOVA_VA;
+               }
+       }
+       RTE_LOG(DEBUG, EAL, "Selected IOVA mode '%s'\n",
+               iova_mode == RTE_IOVA_PA ? "PA" : "VA");
+       rte_eal_get_configuration()->iova_mode = iova_mode;
+
        if (rte_eal_memzone_init() < 0) {
                rte_eal_init_alert("Cannot init memzone");
                rte_errno = ENODEV;
@@ -358,27 +401,9 @@ rte_eal_init(int argc, char **argv)
                return -1;
        }
 
-       if (rte_eal_intr_init() < 0) {
-               rte_eal_init_alert("Cannot init interrupt-handling thread");
-               return -1;
-       }
-
-       if (rte_eal_timer_init() < 0) {
-               rte_eal_init_alert("Cannot init TSC timer");
-               rte_errno = EFAULT;
-               return -1;
-       }
-
        __rte_thread_init(config->main_lcore,
                &lcore_config[config->main_lcore].cpuset);
 
-       bscan = rte_bus_scan();
-       if (bscan < 0) {
-               rte_eal_init_alert("Cannot init PCI");
-               rte_errno = ENODEV;
-               return -1;
-       }
-
        RTE_LCORE_FOREACH_WORKER(i) {
 
                /*

diff --git a/lib/eal/windows/eal_memalloc.c b/lib/eal/windows/eal_memalloc.c
index 4459d59..55d6dcc 100644 (file)
--- a/lib/eal/windows/eal_memalloc.c
+++ b/lib/eal/windows/eal_memalloc.c
@@ -99,16 +99,11 @@ alloc_seg(struct rte_memseg *ms, void *requested_addr, int socket_id,
         */
        *(volatile int *)addr = *(volatile int *)addr;
 
-       /* Only try to obtain IOVA if it's available, so that applications
-        * that do not need IOVA can use this allocator.
-        */
-       if (rte_eal_using_phys_addrs()) {
-               iova = rte_mem_virt2iova(addr);
-               if (iova == RTE_BAD_IOVA) {
-                       RTE_LOG(DEBUG, EAL,
-                               "Cannot get IOVA of allocated segment\n");
-                       goto error;
-               }
+       iova = rte_mem_virt2iova(addr);
+       if (iova == RTE_BAD_IOVA) {
+               RTE_LOG(DEBUG, EAL,
+                       "Cannot get IOVA of allocated segment\n");
+               goto error;
        }
 
        /* Only "Ex" function can handle hugepages. */

diff --git a/lib/eal/windows/eal_memory.c b/lib/eal/windows/eal_memory.c
index 71741fc..2fd37d9 100644 (file)
--- a/lib/eal/windows/eal_memory.c
+++ b/lib/eal/windows/eal_memory.c
@@ -225,19 +225,17 @@ rte_mem_virt2phy(const void *virt)
        return phys.QuadPart;
 }
 
-/* Windows currently only supports IOVA as PA. */
 rte_iova_t
 rte_mem_virt2iova(const void *virt)
 {
        phys_addr_t phys;
 
-       if (virt2phys_device == INVALID_HANDLE_VALUE)
-               return RTE_BAD_IOVA;
+       if (rte_eal_iova_mode() == RTE_IOVA_VA)
+               return (rte_iova_t)virt;
 
        phys = rte_mem_virt2phy(virt);
        if (phys == RTE_BAD_PHYS_ADDR)
                return RTE_BAD_IOVA;
-
        return (rte_iova_t)phys;
 }