[v2,4/5] mem: use memfd for no-huge mode
Checks
Commit Message
When running in no-huge mode, we anonymously allocate our memory.
While this works for regular NICs and vdev's, it's not suitable
for memory sharing scenarios such as virtio with vhost_user
backend.
To fix this, allocate no-huge memory using memfd, and register
it with memalloc just like any other memseg fd. This will enable
using rte_memseg_get_fd() API with --no-huge EAL flag.
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
---
Notes:
v2:
- Detect memfd support at compile time
- Change memfd-related log level to debug
doc/guides/rel_notes/release_19_02.rst | 5 +++
lib/librte_eal/linuxapp/eal/eal_memory.c | 54 +++++++++++++++++++++++-
2 files changed, 57 insertions(+), 2 deletions(-)
Comments
On Tue, Dec 11, 2018 at 04:43:31PM +0000, Anatoly Burakov wrote:
> When running in no-huge mode, we anonymously allocate our memory.
> While this works for regular NICs and vdev's, it's not suitable
> for memory sharing scenarios such as virtio with vhost_user
> backend.
>
> To fix this, allocate no-huge memory using memfd, and register
> it with memalloc just like any other memseg fd. This will enable
> using rte_memseg_get_fd() API with --no-huge EAL flag.
>
> Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
> ---
>
> Notes:
> v2:
> - Detect memfd support at compile time
> - Change memfd-related log level to debug
>
> doc/guides/rel_notes/release_19_02.rst | 5 +++
> lib/librte_eal/linuxapp/eal/eal_memory.c | 54 +++++++++++++++++++++++-
> 2 files changed, 57 insertions(+), 2 deletions(-)
>
> diff --git a/doc/guides/rel_notes/release_19_02.rst b/doc/guides/rel_notes/release_19_02.rst
> index 960098582..420d51b5b 100644
> --- a/doc/guides/rel_notes/release_19_02.rst
> +++ b/doc/guides/rel_notes/release_19_02.rst
> @@ -23,6 +23,11 @@ DPDK Release 19.02
> New Features
> ------------
>
> +* **Support for using VirtIO without hugepages**
> +
> + The --no-huge mode was augmented to use memfd-backed memory (on systems that
> + support memfd), to allow using VirtIO-based NICs without hugepages.
It would be better to say virtio-user here, because virtio NICs
e.g. the one emulated by QEMU, could be something quite different.
> +
> .. This section should contain new features added in this release.
> Sample format:
>
[...]
On 13-Dec-18 4:59 AM, Tiwei Bie wrote:
> On Tue, Dec 11, 2018 at 04:43:31PM +0000, Anatoly Burakov wrote:
>> When running in no-huge mode, we anonymously allocate our memory.
>> While this works for regular NICs and vdev's, it's not suitable
>> for memory sharing scenarios such as virtio with vhost_user
>> backend.
>>
>> To fix this, allocate no-huge memory using memfd, and register
>> it with memalloc just like any other memseg fd. This will enable
>> using rte_memseg_get_fd() API with --no-huge EAL flag.
>>
>> Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
>> ---
>>
>> Notes:
>> v2:
>> - Detect memfd support at compile time
>> - Change memfd-related log level to debug
>>
>> doc/guides/rel_notes/release_19_02.rst | 5 +++
>> lib/librte_eal/linuxapp/eal/eal_memory.c | 54 +++++++++++++++++++++++-
>> 2 files changed, 57 insertions(+), 2 deletions(-)
>>
>> diff --git a/doc/guides/rel_notes/release_19_02.rst b/doc/guides/rel_notes/release_19_02.rst
>> index 960098582..420d51b5b 100644
>> --- a/doc/guides/rel_notes/release_19_02.rst
>> +++ b/doc/guides/rel_notes/release_19_02.rst
>> @@ -23,6 +23,11 @@ DPDK Release 19.02
>> New Features
>> ------------
>>
>> +* **Support for using VirtIO without hugepages**
>> +
>> + The --no-huge mode was augmented to use memfd-backed memory (on systems that
>> + support memfd), to allow using VirtIO-based NICs without hugepages.
>
> It would be better to say virtio-user here, because virtio NICs
> e.g. the one emulated by QEMU, could be something quite different.
Thanks, will fix!
>
>> +
>> .. This section should contain new features added in this release.
>> Sample format:
>>
> [...]
>
@@ -23,6 +23,11 @@ DPDK Release 19.02
New Features
------------
+* **Support for using VirtIO without hugepages**
+
+ The --no-huge mode was augmented to use memfd-backed memory (on systems that
+ support memfd), to allow using VirtIO-based NICs without hugepages.
+
.. This section should contain new features added in this release.
Sample format:
@@ -25,6 +25,10 @@
#include <sys/time.h>
#include <signal.h>
#include <setjmp.h>
+#ifdef F_ADD_SEALS /* if file sealing is supported, so is memfd */
+#include <linux/memfd.h>
+#define MEMFD_SUPPORTED
+#endif
#ifdef RTE_EAL_NUMA_AWARE_HUGEPAGES
#include <numa.h>
#include <numaif.h>
@@ -1341,12 +1345,18 @@ eal_legacy_hugepage_init(void)
/* hugetlbfs can be disabled */
if (internal_config.no_hugetlbfs) {
struct rte_memseg_list *msl;
+ int n_segs, cur_seg, fd, flags;
+#ifdef MEMFD_SUPPORTED
+ int memfd;
+#endif
uint64_t page_sz;
- int n_segs, cur_seg;
/* nohuge mode is legacy mode */
internal_config.legacy_mem = 1;
+ /* nohuge mode is single-file segments mode */
+ internal_config.single_file_segments = 1;
+
/* create a memseg list */
msl = &mcfg->memsegs[0];
@@ -1359,8 +1369,38 @@ eal_legacy_hugepage_init(void)
return -1;
}
+ /* set up parameters for anonymous mmap */
+ fd = -1;
+ flags = MAP_PRIVATE | MAP_ANONYMOUS;
+
+#ifdef MEMFD_SUPPORTED
+ /* create a memfd and store it in the segment fd table */
+ memfd = memfd_create("nohuge", 0);
+ if (memfd < 0) {
+ RTE_LOG(DEBUG, EAL, "Cannot create memfd: %s\n",
+ strerror(errno));
+ RTE_LOG(DEBUG, EAL, "Falling back to anonymous map\n");
+ } else {
+ /* we got an fd - now resize it */
+ if (ftruncate(memfd, internal_config.memory) < 0) {
+ RTE_LOG(ERR, EAL, "Cannot resize memfd: %s\n",
+ strerror(errno));
+ RTE_LOG(ERR, EAL, "Falling back to anonymous map\n");
+ close(memfd);
+ } else {
+ /* creating memfd-backed file was successful.
+ * we want changes to memfd to be visible to
+ * other processes (such as vhost backend), so
+ * map it as shared memory.
+ */
+ RTE_LOG(DEBUG, EAL, "Using memfd for anonymous memory\n");
+ fd = memfd;
+ flags = MAP_SHARED;
+ }
+ }
+#endif
addr = mmap(NULL, internal_config.memory, PROT_READ | PROT_WRITE,
- MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+ flags, fd, 0);
if (addr == MAP_FAILED) {
RTE_LOG(ERR, EAL, "%s: mmap() failed: %s\n", __func__,
strerror(errno));
@@ -1371,6 +1411,16 @@ eal_legacy_hugepage_init(void)
msl->socket_id = 0;
msl->len = internal_config.memory;
+ /* we're in single-file segments mode, so only the segment list
+ * fd needs to be set up.
+ */
+ if (fd != -1) {
+ if (eal_memalloc_set_seg_list_fd(0, fd) < 0) {
+ RTE_LOG(ERR, EAL, "Cannot set up segment list fd\n");
+ /* not a serious error, proceed */
+ }
+ }
+
/* populate memsegs. each memseg is one page long */
for (cur_seg = 0; cur_seg < n_segs; cur_seg++) {
arr = &msl->memseg_arr;