Enrich rdmatool with an option to enable or disable privileged QKEY.
When enabled, non-privileged users will be allowed to specify a
controlled QKEY.
By default this parameter is disabled in order to comply with IB spec.
According to the IB specification rel-1.6, section 3.5.3:
"QKEYs with the most significant bit set are considered controlled
QKEYs, and a HCA does not allow a consumer to arbitrarily specify a
controlled QKEY."
This allows old applications which existed before the kernel commit:
0cadb4db79e1 ("RDMA/uverbs: Restrict usage of privileged QKEYs")
they can use privileged QKEYs without being a privileged user to now
be able to work again without being privileged granted they turn on this
parameter.
rdma tool command examples and output.
$ rdma system show
netns shared privileged-qkey off copy-on-fork on
$ rdma system set privileged-qkey on
$ rdma system show
netns shared privileged-qkey on copy-on-fork on
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Reviewed-by: Michael Guralnik <michaelgur@nvidia.com>
---
rdma/sys.c | 64 ++++++++++++++++++++++++++++++++++++++++++++++++++--
rdma/utils.c | 1 +
2 files changed, 63 insertions(+), 2 deletions(-)
diff --git a/rdma/sys.c b/rdma/sys.c
index fd785b25..32ca3444 100644
--- a/rdma/sys.c
+++ b/rdma/sys.c
@@ -17,6 +17,11 @@ static const char *netns_modes_str[] = {
"shared",
};
+static const char *privileged_qkey_str[] = {
+ "off",
+ "on",
+};
+
static int sys_show_parse_cb(const struct nlmsghdr *nlh, void *data)
{
struct nlattr *tb[RDMA_NLDEV_ATTR_MAX] = {};
@@ -40,6 +45,22 @@ static int sys_show_parse_cb(const struct nlmsghdr *nlh, void *data)
mode_str);
}
+ if (tb[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE]) {
+ const char *pqkey_str;
+ uint8_t pqkey_mode;
+
+ pqkey_mode =
+ mnl_attr_get_u8(tb[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE]);
+
+ if (pqkey_mode < ARRAY_SIZE(privileged_qkey_str))
+ pqkey_str = privileged_qkey_str[pqkey_mode];
+ else
+ pqkey_str = "unknown";
+
+ print_color_string(PRINT_ANY, COLOR_NONE, "privileged-qkey",
+ "privileged-qkey %s ", pqkey_str);
+ }
+
if (tb[RDMA_NLDEV_SYS_ATTR_COPY_ON_FORK])
cof = mnl_attr_get_u8(tb[RDMA_NLDEV_SYS_ATTR_COPY_ON_FORK]);
@@ -67,8 +88,9 @@ static int sys_show_no_args(struct rd *rd)
static int sys_show(struct rd *rd)
{
const struct rd_cmd cmds[] = {
- { NULL, sys_show_no_args},
- { "netns", sys_show_no_args},
+ { NULL, sys_show_no_args},
+ { "netns", sys_show_no_args},
+ { "privileged-qkey", sys_show_no_args},
{ 0 }
};
@@ -86,6 +108,17 @@ static int sys_set_netns_cmd(struct rd *rd, bool enable)
return rd_sendrecv_msg(rd, seq);
}
+static int sys_set_privileged_qkey_cmd(struct rd *rd, bool enable)
+{
+ uint32_t seq;
+
+ rd_prepare_msg(rd, RDMA_NLDEV_CMD_SYS_SET,
+ &seq, (NLM_F_REQUEST | NLM_F_ACK));
+ mnl_attr_put_u8(rd->nlh, RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE, enable);
+
+ return rd_sendrecv_msg(rd, seq);
+}
+
static bool sys_valid_netns_cmd(const char *cmd)
{
int i;
@@ -97,6 +130,17 @@ static bool sys_valid_netns_cmd(const char *cmd)
return false;
}
+static bool sys_valid_privileged_qkey_cmd(const char *cmd)
+{
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(privileged_qkey_str); i++) {
+ if (!strcmp(cmd, privileged_qkey_str[i]))
+ return true;
+ }
+ return false;
+}
+
static int sys_set_netns_args(struct rd *rd)
{
bool cmd;
@@ -111,10 +155,25 @@ static int sys_set_netns_args(struct rd *rd)
return sys_set_netns_cmd(rd, cmd);
}
+static int sys_set_privileged_qkey_args(struct rd *rd)
+{
+ bool cmd;
+
+ if (rd_no_arg(rd) || !sys_valid_privileged_qkey_cmd(rd_argv(rd))) {
+ pr_err("valid options are: { on | off }\n");
+ return -EINVAL;
+ }
+
+ cmd = (strcmp(rd_argv(rd), "on") == 0) ? true : false;
+
+ return sys_set_privileged_qkey_cmd(rd, cmd);
+}
+
static int sys_set_help(struct rd *rd)
{
pr_out("Usage: %s system set [PARAM] value\n", rd->filename);
pr_out(" system set netns { shared | exclusive }\n");
+ pr_out(" system set privileged-qkey { on | off }\n");
return 0;
}
@@ -124,6 +183,7 @@ static int sys_set(struct rd *rd)
{ NULL, sys_set_help },
{ "help", sys_set_help },
{ "netns", sys_set_netns_args},
+ { "privileged-qkey", sys_set_privileged_qkey_args},
{ 0 }
};
diff --git a/rdma/utils.c b/rdma/utils.c
index 8a091c05..09985069 100644
--- a/rdma/utils.c
+++ b/rdma/utils.c
@@ -473,6 +473,7 @@ static const enum mnl_attr_data_type nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
[RDMA_NLDEV_ATTR_STAT_AUTO_MODE_MASK] = MNL_TYPE_U32,
[RDMA_NLDEV_ATTR_DEV_DIM] = MNL_TYPE_U8,
[RDMA_NLDEV_ATTR_RES_RAW] = MNL_TYPE_BINARY,
+ [RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE] = MNL_TYPE_U8,
};
static int rd_attr_check(const struct nlattr *attr, int *typep)
--
2.18.1
Patrisious Haddad <phaddad@nvidia.com> writes:
> @@ -40,6 +45,22 @@ static int sys_show_parse_cb(const struct nlmsghdr *nlh, void *data)
> mode_str);
> }
>
> + if (tb[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE]) {
> + const char *pqkey_str;
> + uint8_t pqkey_mode;
> +
> + pqkey_mode =
> + mnl_attr_get_u8(tb[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE]);
> +
> + if (pqkey_mode < ARRAY_SIZE(privileged_qkey_str))
> + pqkey_str = privileged_qkey_str[pqkey_mode];
> + else
> + pqkey_str = "unknown";
> +
> + print_color_string(PRINT_ANY, COLOR_NONE, "privileged-qkey",
> + "privileged-qkey %s ", pqkey_str);
> + }
> +
Elsewhere in the file, you just use print_color_on_off(), why not here?
> if (tb[RDMA_NLDEV_SYS_ATTR_COPY_ON_FORK])
> cof = mnl_attr_get_u8(tb[RDMA_NLDEV_SYS_ATTR_COPY_ON_FORK]);
>
> @@ -111,10 +155,25 @@ static int sys_set_netns_args(struct rd *rd)
> return sys_set_netns_cmd(rd, cmd);
> }
>
> +static int sys_set_privileged_qkey_args(struct rd *rd)
> +{
> + bool cmd;
> +
> + if (rd_no_arg(rd) || !sys_valid_privileged_qkey_cmd(rd_argv(rd))) {
> + pr_err("valid options are: { on | off }\n");
> + return -EINVAL;
> + }
This could use parse_on_off().
> +
> + cmd = (strcmp(rd_argv(rd), "on") == 0) ? true : false;
> +
> + return sys_set_privileged_qkey_cmd(rd, cmd);
> +}
> +
> static int sys_set_help(struct rd *rd)
> {
> pr_out("Usage: %s system set [PARAM] value\n", rd->filename);
> pr_out(" system set netns { shared | exclusive }\n");
> + pr_out(" system set privileged-qkey { on | off }\n");
> return 0;
> }
>
> @@ -124,6 +183,7 @@ static int sys_set(struct rd *rd)
> { NULL, sys_set_help },
> { "help", sys_set_help },
> { "netns", sys_set_netns_args},
> + { "privileged-qkey", sys_set_privileged_qkey_args},
> { 0 }
> };
The rest of the code looks sane to me, but I'm not familiar with the
feature.
On 10/19/2023 1:38 PM, Petr Machata wrote:
> Patrisious Haddad <phaddad@nvidia.com> writes:
>
>> @@ -40,6 +45,22 @@ static int sys_show_parse_cb(const struct nlmsghdr *nlh, void *data)
>> mode_str);
>> }
>>
>> + if (tb[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE]) {
>> + const char *pqkey_str;
>> + uint8_t pqkey_mode;
>> +
>> + pqkey_mode =
>> + mnl_attr_get_u8(tb[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE]);
>> +
>> + if (pqkey_mode < ARRAY_SIZE(privileged_qkey_str))
>> + pqkey_str = privileged_qkey_str[pqkey_mode];
>> + else
>> + pqkey_str = "unknown";
>> +
>> + print_color_string(PRINT_ANY, COLOR_NONE, "privileged-qkey",
>> + "privileged-qkey %s ", pqkey_str);
>> + }
>> +
> Elsewhere in the file, you just use print_color_on_off(), why not here?
About this as I previously answered I don't really see a big difference
between it and "print_color_string" but if the
maintainer thinks this is an essential change I can fix and re-send.
Thanks for the review.
>
>> if (tb[RDMA_NLDEV_SYS_ATTR_COPY_ON_FORK])
>> cof = mnl_attr_get_u8(tb[RDMA_NLDEV_SYS_ATTR_COPY_ON_FORK]);
>>
>> @@ -111,10 +155,25 @@ static int sys_set_netns_args(struct rd *rd)
>> return sys_set_netns_cmd(rd, cmd);
>> }
>>
>> +static int sys_set_privileged_qkey_args(struct rd *rd)
>> +{
>> + bool cmd;
>> +
>> + if (rd_no_arg(rd) || !sys_valid_privileged_qkey_cmd(rd_argv(rd))) {
>> + pr_err("valid options are: { on | off }\n");
>> + return -EINVAL;
>> + }
> This could use parse_on_off().
More importantly I looked a bit more into it, and I prefer not to use
it, it would also lead to additional error prints that are not
consistent with what we previously had for this API, so I prefer to keep
it as is , so that the error messages for all arguments of this command
be identical.
>
>> +
>> + cmd = (strcmp(rd_argv(rd), "on") == 0) ? true : false;
>> +
>> + return sys_set_privileged_qkey_cmd(rd, cmd);
>> +}
>> +
>> static int sys_set_help(struct rd *rd)
>> {
>> pr_out("Usage: %s system set [PARAM] value\n", rd->filename);
>> pr_out(" system set netns { shared | exclusive }\n");
>> + pr_out(" system set privileged-qkey { on | off }\n");
>> return 0;
>> }
>>
>> @@ -124,6 +183,7 @@ static int sys_set(struct rd *rd)
>> { NULL, sys_set_help },
>> { "help", sys_set_help },
>> { "netns", sys_set_netns_args},
>> + { "privileged-qkey", sys_set_privileged_qkey_args},
>> { 0 }
>> };
> The rest of the code looks sane to me, but I'm not familiar with the
> feature.
On 10/19/2023 1:38 PM, Petr Machata wrote:
> Patrisious Haddad <phaddad@nvidia.com> writes:
>
>> @@ -40,6 +45,22 @@ static int sys_show_parse_cb(const struct nlmsghdr *nlh, void *data)
>> mode_str);
>> }
>>
>> + if (tb[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE]) {
>> + const char *pqkey_str;
>> + uint8_t pqkey_mode;
>> +
>> + pqkey_mode =
>> + mnl_attr_get_u8(tb[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE]);
>> +
>> + if (pqkey_mode < ARRAY_SIZE(privileged_qkey_str))
>> + pqkey_str = privileged_qkey_str[pqkey_mode];
>> + else
>> + pqkey_str = "unknown";
>> +
>> + print_color_string(PRINT_ANY, COLOR_NONE, "privileged-qkey",
>> + "privileged-qkey %s ", pqkey_str);
>> + }
>> +
> Elsewhere in the file, you just use print_color_on_off(), why not here?
The print_color_on_off was used for copy-on-fork which as you see has no
set function,
I was simply trying to be consistent with this file convention & style,
whereas print_color_string was used for the other configurable value
("netns"), I can obviously change that if you all see it as necessary.
>
>> if (tb[RDMA_NLDEV_SYS_ATTR_COPY_ON_FORK])
>> cof = mnl_attr_get_u8(tb[RDMA_NLDEV_SYS_ATTR_COPY_ON_FORK]);
>>
>> @@ -111,10 +155,25 @@ static int sys_set_netns_args(struct rd *rd)
>> return sys_set_netns_cmd(rd, cmd);
>> }
>>
>> +static int sys_set_privileged_qkey_args(struct rd *rd)
>> +{
>> + bool cmd;
>> +
>> + if (rd_no_arg(rd) || !sys_valid_privileged_qkey_cmd(rd_argv(rd))) {
>> + pr_err("valid options are: { on | off }\n");
>> + return -EINVAL;
>> + }
> This could use parse_on_off().
You are absolutely correct, but just as well was trying to maintain same
code style as the previous configurable value we have here, but I think
using parse_on_off here can save us some code.
>
>> +
>> + cmd = (strcmp(rd_argv(rd), "on") == 0) ? true : false;
>> +
>> + return sys_set_privileged_qkey_cmd(rd, cmd);
>> +}
>> +
>> static int sys_set_help(struct rd *rd)
>> {
>> pr_out("Usage: %s system set [PARAM] value\n", rd->filename);
>> pr_out(" system set netns { shared | exclusive }\n");
>> + pr_out(" system set privileged-qkey { on | off }\n");
>> return 0;
>> }
>>
>> @@ -124,6 +183,7 @@ static int sys_set(struct rd *rd)
>> { NULL, sys_set_help },
>> { "help", sys_set_help },
>> { "netns", sys_set_netns_args},
>> + { "privileged-qkey", sys_set_privileged_qkey_args},
>> { 0 }
>> };
> The rest of the code looks sane to me, but I'm not familiar with the
> feature.
If no one else has any comments soon, and these two comments are
actually considered critical I can re-send my patches with those issues
fixed.
On 10/22/23 1:41 AM, Patrisious Haddad wrote:
>
> On 10/19/2023 1:38 PM, Petr Machata wrote:
>> Patrisious Haddad <phaddad@nvidia.com> writes:
>>
>>> @@ -40,6 +45,22 @@ static int sys_show_parse_cb(const struct nlmsghdr
>>> *nlh, void *data)
>>> mode_str);
>>> }
>>> + if (tb[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE]) {
>>> + const char *pqkey_str;
>>> + uint8_t pqkey_mode;
>>> +
>>> + pqkey_mode =
>>> +
>>> mnl_attr_get_u8(tb[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE]);
>>> +
>>> + if (pqkey_mode < ARRAY_SIZE(privileged_qkey_str))
>>> + pqkey_str = privileged_qkey_str[pqkey_mode];
>>> + else
>>> + pqkey_str = "unknown";
>>> +
>>> + print_color_string(PRINT_ANY, COLOR_NONE, "privileged-qkey",
>>> + "privileged-qkey %s ", pqkey_str);
>>> + }
>>> +
>> Elsewhere in the file, you just use print_color_on_off(), why not here?
>
> The print_color_on_off was used for copy-on-fork which as you see has no
> set function,
>
> I was simply trying to be consistent with this file convention & style,
> whereas print_color_string was used for the other configurable value
> ("netns"), I can obviously change that if you all see it as necessary.
>
>>
>>> if (tb[RDMA_NLDEV_SYS_ATTR_COPY_ON_FORK])
>>> cof = mnl_attr_get_u8(tb[RDMA_NLDEV_SYS_ATTR_COPY_ON_FORK]);
>>> @@ -111,10 +155,25 @@ static int sys_set_netns_args(struct rd *rd)
>>> return sys_set_netns_cmd(rd, cmd);
>>> }
>>> +static int sys_set_privileged_qkey_args(struct rd *rd)
>>> +{
>>> + bool cmd;
>>> +
>>> + if (rd_no_arg(rd) || !sys_valid_privileged_qkey_cmd(rd_argv(rd))) {
>>> + pr_err("valid options are: { on | off }\n");
>>> + return -EINVAL;
>>> + }
>> This could use parse_on_off().
> You are absolutely correct, but just as well was trying to maintain same
> code style as the previous configurable value we have here, but I think
> using parse_on_off here can save us some code.
>>
>>> +
>>> + cmd = (strcmp(rd_argv(rd), "on") == 0) ? true : false;
>>> +
>>> + return sys_set_privileged_qkey_cmd(rd, cmd);
>>> +}
>>> +
>>> static int sys_set_help(struct rd *rd)
>>> {
>>> pr_out("Usage: %s system set [PARAM] value\n", rd->filename);
>>> pr_out(" system set netns { shared | exclusive }\n");
>>> + pr_out(" system set privileged-qkey { on | off }\n");
>>> return 0;
>>> }
>>> @@ -124,6 +183,7 @@ static int sys_set(struct rd *rd)
>>> { NULL, sys_set_help },
>>> { "help", sys_set_help },
>>> { "netns", sys_set_netns_args},
>>> + { "privileged-qkey", sys_set_privileged_qkey_args},
>>> { 0 }
>>> };
>> The rest of the code looks sane to me, but I'm not familiar with the
>> feature.
> If no one else has any comments soon, and these two comments are
> actually considered critical I can re-send my patches with those issues
> fixed.
tools packaged with iproute2 should use common code where possible.
On 10/22/2023 7:48 PM, David Ahern wrote:
> External email: Use caution opening links or attachments
>
>
> On 10/22/23 1:41 AM, Patrisious Haddad wrote:
>> On 10/19/2023 1:38 PM, Petr Machata wrote:
>>> Patrisious Haddad <phaddad@nvidia.com> writes:
>>>
>>>> @@ -40,6 +45,22 @@ static int sys_show_parse_cb(const struct nlmsghdr
>>>> *nlh, void *data)
>>>> mode_str);
>>>> }
>>>> + if (tb[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE]) {
>>>> + const char *pqkey_str;
>>>> + uint8_t pqkey_mode;
>>>> +
>>>> + pqkey_mode =
>>>> +
>>>> mnl_attr_get_u8(tb[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE]);
>>>> +
>>>> + if (pqkey_mode < ARRAY_SIZE(privileged_qkey_str))
>>>> + pqkey_str = privileged_qkey_str[pqkey_mode];
>>>> + else
>>>> + pqkey_str = "unknown";
>>>> +
>>>> + print_color_string(PRINT_ANY, COLOR_NONE, "privileged-qkey",
>>>> + "privileged-qkey %s ", pqkey_str);
>>>> + }
>>>> +
>>> Elsewhere in the file, you just use print_color_on_off(), why not here?
>> The print_color_on_off was used for copy-on-fork which as you see has no
>> set function,
>>
>> I was simply trying to be consistent with this file convention & style,
>> whereas print_color_string was used for the other configurable value
>> ("netns"), I can obviously change that if you all see it as necessary.
>>
>>>> if (tb[RDMA_NLDEV_SYS_ATTR_COPY_ON_FORK])
>>>> cof = mnl_attr_get_u8(tb[RDMA_NLDEV_SYS_ATTR_COPY_ON_FORK]);
>>>> @@ -111,10 +155,25 @@ static int sys_set_netns_args(struct rd *rd)
>>>> return sys_set_netns_cmd(rd, cmd);
>>>> }
>>>> +static int sys_set_privileged_qkey_args(struct rd *rd)
>>>> +{
>>>> + bool cmd;
>>>> +
>>>> + if (rd_no_arg(rd) || !sys_valid_privileged_qkey_cmd(rd_argv(rd))) {
>>>> + pr_err("valid options are: { on | off }\n");
>>>> + return -EINVAL;
>>>> + }
>>> This could use parse_on_off().
>> You are absolutely correct, but just as well was trying to maintain same
>> code style as the previous configurable value we have here, but I think
>> using parse_on_off here can save us some code.
>>>> +
>>>> + cmd = (strcmp(rd_argv(rd), "on") == 0) ? true : false;
>>>> +
>>>> + return sys_set_privileged_qkey_cmd(rd, cmd);
>>>> +}
>>>> +
>>>> static int sys_set_help(struct rd *rd)
>>>> {
>>>> pr_out("Usage: %s system set [PARAM] value\n", rd->filename);
>>>> pr_out(" system set netns { shared | exclusive }\n");
>>>> + pr_out(" system set privileged-qkey { on | off }\n");
>>>> return 0;
>>>> }
>>>> @@ -124,6 +183,7 @@ static int sys_set(struct rd *rd)
>>>> { NULL, sys_set_help },
>>>> { "help", sys_set_help },
>>>> { "netns", sys_set_netns_args},
>>>> + { "privileged-qkey", sys_set_privileged_qkey_args},
>>>> { 0 }
>>>> };
>>> The rest of the code looks sane to me, but I'm not familiar with the
>>> feature.
>> If no one else has any comments soon, and these two comments are
>> actually considered critical I can re-send my patches with those issues
>> fixed.
> tools packaged with iproute2 should use common code where possible.
Okay, good point, fixed both comments and sent a V2 , it actually
resulted in a much cleaner code.
- Thanks.
On 10/19/23 4:38 AM, Petr Machata wrote:
>
> Patrisious Haddad <phaddad@nvidia.com> writes:
>
>> @@ -40,6 +45,22 @@ static int sys_show_parse_cb(const struct nlmsghdr *nlh, void *data)
>> mode_str);
>> }
>>
>> + if (tb[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE]) {
>> + const char *pqkey_str;
>> + uint8_t pqkey_mode;
>> +
>> + pqkey_mode =
>> + mnl_attr_get_u8(tb[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE]);
>> +
>> + if (pqkey_mode < ARRAY_SIZE(privileged_qkey_str))
>> + pqkey_str = privileged_qkey_str[pqkey_mode];
>> + else
>> + pqkey_str = "unknown";
>> +
>> + print_color_string(PRINT_ANY, COLOR_NONE, "privileged-qkey",
>> + "privileged-qkey %s ", pqkey_str);
>> + }
>> +
>
> Elsewhere in the file, you just use print_color_on_off(), why not here?
+1
© 2016 - 2026 Red Hat, Inc.