[v4] net/gve: check driver compatibility
Checks
Commit Message
Change gve_driver_info fields to report DPDK as OS type and DPDK RTE
version as OS version, reserving driver_version fields for GVE driver
version based on features.
Signed-off-by: Rushil Gupta <rushilg@google.com>
Signed-off-by: Joshua Washington <joshwash@google.com>
Signed-off-by: Junfeng Guo <junfeng.guo@intel.com>
Signed-off-by: Jeroen de Borst <jeroendb@google.com>
---
drivers/net/gve/base/gve.h | 3 --
drivers/net/gve/base/gve_adminq.c | 19 +++++++++
drivers/net/gve/base/gve_adminq.h | 46 ++++++++++++++++++++-
drivers/net/gve/base/gve_osdep.h | 24 +++++++++++
drivers/net/gve/gve_ethdev.c | 67 ++++++++++++++++++++++++++-----
drivers/net/gve/gve_ethdev.h | 2 +-
drivers/net/gve/gve_version.c | 13 ++++++
drivers/net/gve/gve_version.h | 24 +++++++++++
drivers/net/gve/meson.build | 5 ++-
9 files changed, 185 insertions(+), 18 deletions(-)
create mode 100644 drivers/net/gve/gve_version.c
create mode 100644 drivers/net/gve/gve_version.h
Comments
On Fri, 19 May 2023 13:46:18 -0700
Rushil Gupta <rushilg@google.com> wrote:
> +#include <rte_version.h>
>
> #include "../gve_logs.h"
>
> +#ifdef __linux__
> +#include <sys/utsname.h>
> +#endif
> +
> typedef uint8_t u8;
> typedef uint16_t u16;
> typedef uint32_t u32;
> @@ -73,6 +78,12 @@ typedef rte_iova_t dma_addr_t;
>
> #define msleep(ms) rte_delay_ms(ms)
>
> +#define OS_VERSION_STRLEN 128
> +struct os_version_string {
> + char os_version_str1[OS_VERSION_STRLEN];
> + char os_version_str2[OS_VERSION_STRLEN];
> +};
> +
Not sure this a good idea. Are you having the host validate
against DPDK versions. This is a bad idea.
Better to use feature bits like virtio and not be creating
and validating strings about versions. For example, ever minor
stable release changes this.
We are not validating anything.
This is for our internal analysis and product requirements.
On Fri, May 19, 2023 at 1:56 PM Stephen Hemminger <
stephen@networkplumber.org> wrote:
> On Fri, 19 May 2023 13:46:18 -0700
> Rushil Gupta <rushilg@google.com> wrote:
>
> > +#include <rte_version.h>
> >
> > #include "../gve_logs.h"
> >
> > +#ifdef __linux__
> > +#include <sys/utsname.h>
> > +#endif
> > +
> > typedef uint8_t u8;
> > typedef uint16_t u16;
> > typedef uint32_t u32;
> > @@ -73,6 +78,12 @@ typedef rte_iova_t dma_addr_t;
> >
> > #define msleep(ms) rte_delay_ms(ms)
> >
> > +#define OS_VERSION_STRLEN 128
> > +struct os_version_string {
> > + char os_version_str1[OS_VERSION_STRLEN];
> > + char os_version_str2[OS_VERSION_STRLEN];
> > +};
> > +
>
> Not sure this a good idea. Are you having the host validate
> against DPDK versions. This is a bad idea.
>
> Better to use feature bits like virtio and not be creating
> and validating strings about versions. For example, ever minor
> stable release changes this.
>
On 5/19/2023 9:46 PM, Rushil Gupta wrote:
> +static int
> +gve_verify_driver_compatibility(struct gve_priv *priv)
> +{
> + const struct rte_memzone *driver_info_mem;
> + struct gve_driver_info *driver_info;
> + int err;
> +
> + driver_info_mem = rte_memzone_reserve_aligned("verify_driver_compatibility",
> + sizeof(struct gve_driver_info),
> + rte_socket_id(),
> + RTE_MEMZONE_IOVA_CONTIG, PAGE_SIZE);
> +
> + if (driver_info_mem == NULL) {
> + PMD_DRV_LOG(ERR,
> + "Could not alloc memzone for driver compatibility");
> + return -ENOMEM;
> + }
> + driver_info = (struct gve_driver_info *)driver_info_mem->addr;
> +
> + *driver_info = (struct gve_driver_info) {
> + .os_type = 5, /* DPDK */
> + .driver_major = GVE_VERSION_MAJOR,
> + .driver_minor = GVE_VERSION_MINOR,
> + .driver_sub = GVE_VERSION_SUB,
> + .os_version_major = cpu_to_be32(DPDK_VERSION_MAJOR),
> + .os_version_minor = cpu_to_be32(DPDK_VERSION_MINOR),
> + .os_version_sub = cpu_to_be32(DPDK_VERSION_SUB),
> + .driver_capability_flags = {
> + cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS1),
> + cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS2),
> + cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS3),
> + cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS4),
> + },
> + };
> +
> + populate_driver_version_strings((char *)driver_info->os_version_str1,
> + (char *)driver_info->os_version_str2);
> +
> + err = gve_adminq_verify_driver_compatibility(priv,
> + sizeof(struct gve_driver_info), (dma_addr_t)driver_info);
Back to previous discussion, other commands pass physical address to the
admin command, but this pass virtual address.
To follow the same semantic, shouldn't above be 'driver_info_mem.iova'?
I asked before but not able to get an answer, what is the memory type
requirement for device?
Why virtual address obtained via 'calloc()' is not working, but virtual
address from hugepages are working?
1. This is the excerpt from the google's virtual nic spec:
"In addition to the device-owned register file, vector table, and
doorbells, the gVNIC device uses *DMA* (which in most cases amounts to
ordinary memory access by host software since we're dealing with a virtual
device, but guests must assume the device could be backed by actual
hardware) to access physical memory. The following are all located in
physical memory: Admin queue - 4096-byte command queue used for configuring
gVNIC.
Some commands require an additional dma memory region to be passed to the
device. These memory regions are allocated to execute the command and freed
when the command completes."
The calloc by default doesn't allow memory to be shared between the dpdk
process and hypervisor (where virtual device lives); so that's the reason
it doesn't work.
2. I also have a query: RHEL8 compilation in ci/Intel-compilation context
fails due to; is this because of if `not is_linux`
meson.build:67:0: ERROR: Include dir lib/eal/linux/include does not exist.
Passes: http://patchwork.dpdk.org/project/dpdk/patch/20230508191552.104540-1-rushilg@google.com/
Fails: http://patchwork.dpdk.org/project/dpdk/patch/20230519204618.1507956-1-rushilg@google.com/
On Mon, May 22, 2023 at 1:52 AM Ferruh Yigit <ferruh.yigit@amd.com> wrote:
> On 5/19/2023 9:46 PM, Rushil Gupta wrote:
> > +static int
> > +gve_verify_driver_compatibility(struct gve_priv *priv)
> > +{
> > + const struct rte_memzone *driver_info_mem;
> > + struct gve_driver_info *driver_info;
> > + int err;
> > +
> > + driver_info_mem =
> rte_memzone_reserve_aligned("verify_driver_compatibility",
> > + sizeof(struct gve_driver_info),
> > + rte_socket_id(),
> > + RTE_MEMZONE_IOVA_CONTIG, PAGE_SIZE);
> > +
> > + if (driver_info_mem == NULL) {
> > + PMD_DRV_LOG(ERR,
> > + "Could not alloc memzone for driver
> compatibility");
> > + return -ENOMEM;
> > + }
> > + driver_info = (struct gve_driver_info *)driver_info_mem->addr;
> > +
> > + *driver_info = (struct gve_driver_info) {
> > + .os_type = 5, /* DPDK */
> > + .driver_major = GVE_VERSION_MAJOR,
> > + .driver_minor = GVE_VERSION_MINOR,
> > + .driver_sub = GVE_VERSION_SUB,
> > + .os_version_major = cpu_to_be32(DPDK_VERSION_MAJOR),
> > + .os_version_minor = cpu_to_be32(DPDK_VERSION_MINOR),
> > + .os_version_sub = cpu_to_be32(DPDK_VERSION_SUB),
> > + .driver_capability_flags = {
> > + cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS1),
> > + cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS2),
> > + cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS3),
> > + cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS4),
> > + },
> > + };
> > +
> > + populate_driver_version_strings((char
> *)driver_info->os_version_str1,
> > + (char *)driver_info->os_version_str2);
> > +
> > + err = gve_adminq_verify_driver_compatibility(priv,
> > + sizeof(struct gve_driver_info), (dma_addr_t)driver_info);
>
> Back to previous discussion, other commands pass physical address to the
> admin command, but this pass virtual address.
> To follow the same semantic, shouldn't above be 'driver_info_mem.iova'?
>
> I asked before but not able to get an answer, what is the memory type
> requirement for device?
> Why virtual address obtained via 'calloc()' is not working, but virtual
> address from hugepages are working?
>
On 5/22/2023 4:45 PM, Rushil Gupta wrote:
> 1. This is the excerpt from the google's virtual nic spec:
> "In addition to the device-owned register file, vector table, and
> doorbells, the gVNIC device uses *DMA* (which in most cases amounts to
> ordinary memory access by host software since we're dealing with a
> virtual device, but guests must assume the device could be backed by
> actual hardware) to access physical memory. The following are all
> located in physical memory: Admin queue - 4096-byte command queue used
> for configuring gVNIC.
> Some commands require an additional dma memory region to be passed to
> the device. These memory regions are allocated to execute the command
> and freed when the command completes."
> The calloc by default doesn't allow memory to be shared between the dpdk
> process and hypervisor (where virtual device lives); so that's the
> reason it doesn't work.
>
Thanks Rushil for the info.
So, I expect gVNIC requires physical address to be passed in the admin
command, as 'driver_info_mem.iova'.
What confuses me is, latest version passes another virtual address
'driver_info' ('driver_info_mem->addr').
> 2. I also have a query: RHEL8 compilation in ci/Intel-compilation
> context fails due to; is this because of if `not is_linux`
>
> meson.build:67:0: ERROR: Include dir lib/eal/linux/include does not exist.
>
This error shouldn't be related with `not is_linux`, but I am not sure
about its root case, if it still exists in next version we can
communicate with CI team for details. For now I assume this is an
infrastructure issue.
> Passes:
> http://patchwork.dpdk.org/project/dpdk/patch/20230508191552.104540-1-rushilg@google.com/ <http://patchwork.dpdk.org/project/dpdk/patch/20230508191552.104540-1-rushilg@google.com/>
>
> Fails:
> http://patchwork.dpdk.org/project/dpdk/patch/20230519204618.1507956-1-rushilg@google.com/ <http://patchwork.dpdk.org/project/dpdk/patch/20230519204618.1507956-1-rushilg@google.com/>
>
>
> On Mon, May 22, 2023 at 1:52 AM Ferruh Yigit <ferruh.yigit@amd.com
> <mailto:ferruh.yigit@amd.com>> wrote:
>
> On 5/19/2023 9:46 PM, Rushil Gupta wrote:
> > +static int
> > +gve_verify_driver_compatibility(struct gve_priv *priv)
> > +{
> > + const struct rte_memzone *driver_info_mem;
> > + struct gve_driver_info *driver_info;
> > + int err;
> > +
> > + driver_info_mem =
> rte_memzone_reserve_aligned("verify_driver_compatibility",
> > + sizeof(struct gve_driver_info),
> > + rte_socket_id(),
> > + RTE_MEMZONE_IOVA_CONTIG, PAGE_SIZE);
> > +
> > + if (driver_info_mem == NULL) {
> > + PMD_DRV_LOG(ERR,
> > + "Could not alloc memzone for driver
> compatibility");
> > + return -ENOMEM;
> > + }
> > + driver_info = (struct gve_driver_info *)driver_info_mem->addr;
> > +
> > + *driver_info = (struct gve_driver_info) {
> > + .os_type = 5, /* DPDK */
> > + .driver_major = GVE_VERSION_MAJOR,
> > + .driver_minor = GVE_VERSION_MINOR,
> > + .driver_sub = GVE_VERSION_SUB,
> > + .os_version_major = cpu_to_be32(DPDK_VERSION_MAJOR),
> > + .os_version_minor = cpu_to_be32(DPDK_VERSION_MINOR),
> > + .os_version_sub = cpu_to_be32(DPDK_VERSION_SUB),
> > + .driver_capability_flags = {
> > + cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS1),
> > + cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS2),
> > + cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS3),
> > + cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS4),
> > + },
> > + };
> > +
> > + populate_driver_version_strings((char
> *)driver_info->os_version_str1,
> > + (char *)driver_info->os_version_str2);
> > +
> > + err = gve_adminq_verify_driver_compatibility(priv,
> > + sizeof(struct gve_driver_info),
> (dma_addr_t)driver_info);
>
> Back to previous discussion, other commands pass physical address to the
> admin command, but this pass virtual address.
> To follow the same semantic, shouldn't above be 'driver_info_mem.iova'?
>
> I asked before but not able to get an answer, what is the memory type
> requirement for device?
> Why virtual address obtained via 'calloc()' is not working, but virtual
> address from hugepages are working?
>
On 5/19/2023 9:46 PM, Rushil Gupta wrote:
> diff --git a/drivers/net/gve/base/gve_osdep.h b/drivers/net/gve/base/gve_osdep.h
> index abf3d379ae..5e8ae1eac6 100644
> --- a/drivers/net/gve/base/gve_osdep.h
> +++ b/drivers/net/gve/base/gve_osdep.h
> @@ -21,9 +21,14 @@
> #include <rte_malloc.h>
> #include <rte_memcpy.h>
> #include <rte_memzone.h>
> +#include <rte_version.h>
>
> #include "../gve_logs.h"
>
> +#ifdef __linux__
> +#include <sys/utsname.h>
> +#endif
> +
Can you please use 'RTE_EXEC_ENV_LINUX' macro instead of '__linux__'?
As noted from spec: "Some commands require an additional dma memory
region to be passed to the device"
We are passing virtual addresses to devices (please look at link-speed
and describe-device adminq commands):
describe device:
https://github.com/DPDK/dpdk/blob/main/drivers/net/gve/base/gve_adminq.c#L704
link-speed: https://github.com/DPDK/dpdk/blob/main/drivers/net/gve/base/gve_adminq.c#L869
These adminq commands rely on gve_dma_alloc which presents mz->addr
and not iova as memory region:
https://github.com/DPDK/dpdk/blob/main/drivers/net/gve/base/gve_osdep.h#L139
Here is the new patch:
http://patchwork.dpdk.org/project/dpdk/patch/20230524171324.2072742-1-rushilg@google.com/
On Tue, May 23, 2023 at 3:22 AM Ferruh Yigit <ferruh.yigit@amd.com> wrote:
>
> On 5/19/2023 9:46 PM, Rushil Gupta wrote:
> > diff --git a/drivers/net/gve/base/gve_osdep.h b/drivers/net/gve/base/gve_osdep.h
> > index abf3d379ae..5e8ae1eac6 100644
> > --- a/drivers/net/gve/base/gve_osdep.h
> > +++ b/drivers/net/gve/base/gve_osdep.h
> > @@ -21,9 +21,14 @@
> > #include <rte_malloc.h>
> > #include <rte_memcpy.h>
> > #include <rte_memzone.h>
> > +#include <rte_version.h>
> >
> > #include "../gve_logs.h"
> >
> > +#ifdef __linux__
> > +#include <sys/utsname.h>
> > +#endif
> > +
>
> Can you please use 'RTE_EXEC_ENV_LINUX' macro instead of '__linux__'?
http://patchwork.dpdk.org/project/dpdk/patch/20230524171324.2072742-1-rushilg@google.com/
Does this look good to you?
On Wed, May 24, 2023 at 10:14 AM Rushil Gupta <rushilg@google.com> wrote:
>
> As noted from spec: "Some commands require an additional dma memory
> region to be passed to the device"
> We are passing virtual addresses to devices (please look at link-speed
> and describe-device adminq commands):
> describe device:
> https://github.com/DPDK/dpdk/blob/main/drivers/net/gve/base/gve_adminq.c#L704
> link-speed: https://github.com/DPDK/dpdk/blob/main/drivers/net/gve/base/gve_adminq.c#L869
>
> These adminq commands rely on gve_dma_alloc which presents mz->addr
> and not iova as memory region:
> https://github.com/DPDK/dpdk/blob/main/drivers/net/gve/base/gve_osdep.h#L139
>
> Here is the new patch:
> http://patchwork.dpdk.org/project/dpdk/patch/20230524171324.2072742-1-rushilg@google.com/
>
>
>
> On Tue, May 23, 2023 at 3:22 AM Ferruh Yigit <ferruh.yigit@amd.com> wrote:
> >
> > On 5/19/2023 9:46 PM, Rushil Gupta wrote:
> > > diff --git a/drivers/net/gve/base/gve_osdep.h b/drivers/net/gve/base/gve_osdep.h
> > > index abf3d379ae..5e8ae1eac6 100644
> > > --- a/drivers/net/gve/base/gve_osdep.h
> > > +++ b/drivers/net/gve/base/gve_osdep.h
> > > @@ -21,9 +21,14 @@
> > > #include <rte_malloc.h>
> > > #include <rte_memcpy.h>
> > > #include <rte_memzone.h>
> > > +#include <rte_version.h>
> > >
> > > #include "../gve_logs.h"
> > >
> > > +#ifdef __linux__
> > > +#include <sys/utsname.h>
> > > +#endif
> > > +
> >
> > Can you please use 'RTE_EXEC_ENV_LINUX' macro instead of '__linux__'?
@@ -9,9 +9,6 @@
#include "gve_desc.h"
#include "gve_desc_dqo.h"
-#define GVE_VERSION "1.3.0"
-#define GVE_VERSION_PREFIX "GVE-"
-
#ifndef GOOGLE_VENDOR_ID
#define GOOGLE_VENDOR_ID 0x1ae0
#endif
@@ -401,6 +401,9 @@ static int gve_adminq_issue_cmd(struct gve_priv *priv,
case GVE_ADMINQ_GET_PTYPE_MAP:
priv->adminq_get_ptype_map_cnt++;
break;
+ case GVE_ADMINQ_VERIFY_DRIVER_COMPATIBILITY:
+ priv->adminq_verify_driver_compatibility_cnt++;
+ break;
default:
PMD_DRV_LOG(ERR, "unknown AQ command opcode %d", opcode);
}
@@ -465,6 +468,22 @@ int gve_adminq_configure_device_resources(struct gve_priv *priv,
return gve_adminq_execute_cmd(priv, &cmd);
}
+int gve_adminq_verify_driver_compatibility(struct gve_priv *priv,
+ u64 driver_info_len,
+ dma_addr_t driver_info_addr)
+{
+ union gve_adminq_command cmd;
+
+ memset(&cmd, 0, sizeof(cmd));
+ cmd.opcode = cpu_to_be32(GVE_ADMINQ_VERIFY_DRIVER_COMPATIBILITY);
+ cmd.verify_driver_compatibility = (struct gve_adminq_verify_driver_compatibility) {
+ .driver_info_len = cpu_to_be64(driver_info_len),
+ .driver_info_addr = cpu_to_be64(driver_info_addr),
+ };
+
+ return gve_adminq_execute_cmd(priv, &cmd);
+}
+
int gve_adminq_deconfigure_device_resources(struct gve_priv *priv)
{
union gve_adminq_command cmd;
@@ -1,6 +1,6 @@
/* SPDX-License-Identifier: MIT
* Google Virtual Ethernet (gve) driver
- * Copyright (C) 2015-2022 Google, Inc.
+ * Copyright (C) 2015-2023 Google, Inc.
*/
#ifndef _GVE_ADMINQ_H
@@ -23,6 +23,7 @@ enum gve_adminq_opcodes {
GVE_ADMINQ_REPORT_STATS = 0xC,
GVE_ADMINQ_REPORT_LINK_SPEED = 0xD,
GVE_ADMINQ_GET_PTYPE_MAP = 0xE,
+ GVE_ADMINQ_VERIFY_DRIVER_COMPATIBILITY = 0xF,
};
/* Admin queue status codes */
@@ -145,6 +146,44 @@ enum gve_sup_feature_mask {
};
#define GVE_DEV_OPT_LEN_GQI_RAW_ADDRESSING 0x0
+enum gve_driver_capbility {
+ gve_driver_capability_gqi_qpl = 0,
+ gve_driver_capability_gqi_rda = 1,
+ gve_driver_capability_dqo_qpl = 2, /* reserved for future use */
+ gve_driver_capability_dqo_rda = 3,
+};
+
+#define GVE_CAP1(a) BIT((int)a)
+
+#define GVE_DRIVER_CAPABILITY_FLAGS1 \
+ (GVE_CAP1(gve_driver_capability_gqi_qpl) | \
+ GVE_CAP1(gve_driver_capability_gqi_rda) | \
+ GVE_CAP1(gve_driver_capability_dqo_rda))
+
+#define GVE_DRIVER_CAPABILITY_FLAGS2 0x0
+#define GVE_DRIVER_CAPABILITY_FLAGS3 0x0
+#define GVE_DRIVER_CAPABILITY_FLAGS4 0x0
+
+struct gve_driver_info {
+ u8 os_type; /* 0x05 = DPDK */
+ u8 driver_major;
+ u8 driver_minor;
+ u8 driver_sub;
+ __be32 os_version_major;
+ __be32 os_version_minor;
+ __be32 os_version_sub;
+ __be64 driver_capability_flags[4];
+ u8 os_version_str1[OS_VERSION_STRLEN];
+ u8 os_version_str2[OS_VERSION_STRLEN];
+};
+
+struct gve_adminq_verify_driver_compatibility {
+ __be64 driver_info_len;
+ __be64 driver_info_addr;
+};
+
+GVE_CHECK_STRUCT_LEN(16, gve_adminq_verify_driver_compatibility);
+
struct gve_adminq_configure_device_resources {
__be64 counter_array;
@@ -345,6 +384,8 @@ union gve_adminq_command {
struct gve_adminq_report_stats report_stats;
struct gve_adminq_report_link_speed report_link_speed;
struct gve_adminq_get_ptype_map get_ptype_map;
+ struct gve_adminq_verify_driver_compatibility
+ verify_driver_compatibility;
};
};
u8 reserved[64];
@@ -378,4 +419,7 @@ struct gve_ptype_lut;
int gve_adminq_get_ptype_map_dqo(struct gve_priv *priv,
struct gve_ptype_lut *ptype_lut);
+int gve_adminq_verify_driver_compatibility(struct gve_priv *priv,
+ u64 driver_info_len,
+ dma_addr_t driver_info_addr);
#endif /* _GVE_ADMINQ_H */
@@ -21,9 +21,14 @@
#include <rte_malloc.h>
#include <rte_memcpy.h>
#include <rte_memzone.h>
+#include <rte_version.h>
#include "../gve_logs.h"
+#ifdef __linux__
+#include <sys/utsname.h>
+#endif
+
typedef uint8_t u8;
typedef uint16_t u16;
typedef uint32_t u32;
@@ -73,6 +78,12 @@ typedef rte_iova_t dma_addr_t;
#define msleep(ms) rte_delay_ms(ms)
+#define OS_VERSION_STRLEN 128
+struct os_version_string {
+ char os_version_str1[OS_VERSION_STRLEN];
+ char os_version_str2[OS_VERSION_STRLEN];
+};
+
/* These macros are used to generate compilation errors if a struct/union
* is not exactly the correct length. It gives a divide by zero error if
* the struct/union is not of the correct size, otherwise it creates an
@@ -160,4 +171,17 @@ gve_free_dma_mem(struct gve_dma_mem *mem)
mem->pa = 0;
}
+static inline void
+populate_driver_version_strings(char *str1, char *str2)
+{
+ struct utsname uts;
+ if (uname(&uts) >= 0) {
+ /* release */
+ rte_strscpy(str1, uts.release,
+ OS_VERSION_STRLEN);
+ /* version */
+ rte_strscpy(str2, uts.version,
+ OS_VERSION_STRLEN);
+ }
+}
#endif /* _GVE_OSDEP_H_ */
@@ -5,21 +5,13 @@
#include "gve_ethdev.h"
#include "base/gve_adminq.h"
#include "base/gve_register.h"
-
-const char gve_version_str[] = GVE_VERSION;
-static const char gve_version_prefix[] = GVE_VERSION_PREFIX;
+#include "base/gve_osdep.h"
+#include "gve_version.h"
static void
gve_write_version(uint8_t *driver_version_register)
{
- const char *c = gve_version_prefix;
-
- while (*c) {
- writeb(*c, driver_version_register);
- c++;
- }
-
- c = gve_version_str;
+ const char *c = gve_version_string();
while (*c) {
writeb(*c, driver_version_register);
c++;
@@ -245,6 +237,54 @@ gve_dev_close(struct rte_eth_dev *dev)
return err;
}
+static int
+gve_verify_driver_compatibility(struct gve_priv *priv)
+{
+ const struct rte_memzone *driver_info_mem;
+ struct gve_driver_info *driver_info;
+ int err;
+
+ driver_info_mem = rte_memzone_reserve_aligned("verify_driver_compatibility",
+ sizeof(struct gve_driver_info),
+ rte_socket_id(),
+ RTE_MEMZONE_IOVA_CONTIG, PAGE_SIZE);
+
+ if (driver_info_mem == NULL) {
+ PMD_DRV_LOG(ERR,
+ "Could not alloc memzone for driver compatibility");
+ return -ENOMEM;
+ }
+ driver_info = (struct gve_driver_info *)driver_info_mem->addr;
+
+ *driver_info = (struct gve_driver_info) {
+ .os_type = 5, /* DPDK */
+ .driver_major = GVE_VERSION_MAJOR,
+ .driver_minor = GVE_VERSION_MINOR,
+ .driver_sub = GVE_VERSION_SUB,
+ .os_version_major = cpu_to_be32(DPDK_VERSION_MAJOR),
+ .os_version_minor = cpu_to_be32(DPDK_VERSION_MINOR),
+ .os_version_sub = cpu_to_be32(DPDK_VERSION_SUB),
+ .driver_capability_flags = {
+ cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS1),
+ cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS2),
+ cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS3),
+ cpu_to_be64(GVE_DRIVER_CAPABILITY_FLAGS4),
+ },
+ };
+
+ populate_driver_version_strings((char *)driver_info->os_version_str1,
+ (char *)driver_info->os_version_str2);
+
+ err = gve_adminq_verify_driver_compatibility(priv,
+ sizeof(struct gve_driver_info), (dma_addr_t)driver_info);
+ /* It's ok if the device doesn't support this */
+ if (err == -EOPNOTSUPP)
+ err = 0;
+
+ rte_memzone_free(driver_info_mem);
+ return err;
+}
+
static int
gve_dev_info_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
{
@@ -679,6 +719,11 @@ gve_init_priv(struct gve_priv *priv, bool skip_describe_device)
PMD_DRV_LOG(ERR, "Failed to alloc admin queue: err=%d", err);
return err;
}
+ err = gve_verify_driver_compatibility(priv);
+ if (err) {
+ PMD_DRV_LOG(ERR, "Could not verify driver compatibility: err=%d", err);
+ goto free_adminq;
+ }
if (skip_describe_device)
goto setup_device;
@@ -268,7 +268,7 @@ struct gve_priv {
uint32_t adminq_report_stats_cnt;
uint32_t adminq_report_link_speed_cnt;
uint32_t adminq_get_ptype_map_cnt;
-
+ uint32_t adminq_verify_driver_compatibility_cnt;
volatile uint32_t state_flags;
/* Gvnic device link speed from hypervisor. */
new file mode 100644
@@ -0,0 +1,13 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright (C) 2015-2023 Google, Inc.
+ */
+#include "gve_version.h"
+
+const char *gve_version_string(void)
+{
+ static char gve_version[20];
+ snprintf(gve_version, sizeof(gve_version), "%s%d.%d.%d",
+ GVE_VERSION_PREFIX, GVE_VERSION_MAJOR, GVE_VERSION_MINOR,
+ GVE_VERSION_SUB);
+ return gve_version;
+}
new file mode 100644
@@ -0,0 +1,24 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright (C) 2015-2023 Google, Inc.
+ */
+
+#ifndef _GVE_VERSION_H_
+#define _GVE_VERSION_H_
+
+#include <rte_version.h>
+
+#define GVE_VERSION_PREFIX "DPDK-"
+#define GVE_VERSION_MAJOR 1
+#define GVE_VERSION_MINOR 0
+#define GVE_VERSION_SUB 0
+
+#define DPDK_VERSION_MAJOR (100 * RTE_VER_YEAR + RTE_VER_MONTH)
+#define DPDK_VERSION_MINOR RTE_VER_MINOR
+#define DPDK_VERSION_SUB RTE_VER_RELEASE
+
+
+const char *
+gve_version_string(void);
+
+
+#endif /* GVE_VERSION_H */
@@ -1,9 +1,9 @@
# SPDX-License-Identifier: BSD-3-Clause
# Copyright(C) 2022 Intel Corporation
-if is_windows
+if not is_linux
build = false
- reason = 'not supported on Windows'
+ reason = 'only supported on Linux'
subdir_done()
endif
@@ -14,5 +14,6 @@ sources = files(
'gve_rx_dqo.c',
'gve_tx_dqo.c',
'gve_ethdev.c',
+ 'gve_version.c',
)
includes += include_directories('base')