| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <20240409015350.52377-1-renmingshuai@huawei.com>
Date: Tue, 9 Apr 2024 09:53:50 +0800
From: renmingshuai <renmingshuai@...wei.com>
To: <stephen@...workplumber.org>
CC: <dsahern@...il.com>, <liaichun@...wei.com>, <netdev@...r.kernel.org>,
<renmingshuai@...wei.com>, <yanan@...wei.com>
Subject: Re: [PATCH] iplink: add an option to set IFLA_EXT_MASK attribute
> > >> Kernel has add IFLA_EXT_MASK attribute for indicating that certain
> > >> extended ifinfo values are requested by the user application. The ip
> > >> link show cmd always request VFs extended ifinfo.
> > >>
> > >> RTM_GETLINK for greater than about 220 VFs truncates IFLA_VFINFO_LIST
> > >> due to the maximum reach of nlattr's nla_len being exceeded.
> > >> As a result, ip link show command only show the truncated VFs info
> > >> sucn as:
> > >>
> > >> #ip link show dev eth0
> > >> 1: eth0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 ...
> > >> link/ether ...
> > >> vf 0 link/ether 00:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff
> > >> ...
> > >> Truncated VF list: eth0
> > >>
> > >> Add an option to set IFLA_EXT_MASK attribute and users can choose to
> > >> show the extended ifinfo or not.
> > >>
> > >> Signed-off-by: Mingshuai Ren <renmingshuai@...wei.com>
> > >
> > > Adding a new option with on/off seems like more than is necessary.
> > > If we need an option it should just be one word. Any new filter should
> > > have same conventions as existing filters. Maybe 'novf'
> > >
> > > And it looks like not sending IFLA_EXT_MASK will break the changes
> > > made for the link filter already done for VF's.
> >
> > Thanks for your reply. As you suggested, I've added an option
> > named noVF, which has same conventions as existing filter.
> > Also, this new patch does not send RTEXT_FILTER_VF instead of
> > IFLA_EXT_MASK, and it does not break the changes made for the link
> > filter already done for VF's.
> > Please review it again.
>
> Did you read my comment. Any new option should look the same as all
> the other options in the command. Is there any option to ip commands
> that uses mixed case? No. Please stick to lower case.
Apologies for not considering the use of lower case. I have replaced noVF with novf.
Please review it again. Thanks a lot.
By the way, do I need to submit a new patch?
---
ip/ip_common.h | 1 +
ip/ipaddress.c | 15 ++++++++++-----
ip/iplink.c | 2 +-
man/man8/ip-link.8.in | 7 ++++++-
4 files changed, 18 insertions(+), 7 deletions(-)
diff --git a/ip/ip_common.h b/ip/ip_common.h
index b65c2b41..d3645a2c 100644
--- a/ip/ip_common.h
+++ b/ip/ip_common.h
@@ -30,6 +30,7 @@ struct link_filter {
int target_nsid;
bool have_proto;
int proto;
+ int vfinfo;
};
const char *get_ip_lib_dir(void);
diff --git a/ip/ipaddress.c b/ip/ipaddress.c
index e536912f..e7e5ce76 100644
--- a/ip/ipaddress.c
+++ b/ip/ipaddress.c
@@ -2029,10 +2029,11 @@ static int ipaddr_flush(void)
static int iplink_filter_req(struct nlmsghdr *nlh, int reqlen)
{
- __u32 filt_mask;
+ __u32 filt_mask = 0;
int err;
- filt_mask = RTEXT_FILTER_VF;
+ if (!filter.vfinfo)
+ filt_mask |= RTEXT_FILTER_VF;
if (!show_stats)
filt_mask |= RTEXT_FILTER_SKIP_STATS;
err = addattr32(nlh, reqlen, IFLA_EXT_MASK, filt_mask);
@@ -2070,12 +2071,13 @@ static int ipaddr_link_get(int index, struct nlmsg_chain *linfo)
.i.ifi_family = filter.family,
.i.ifi_index = index,
};
- __u32 filt_mask = RTEXT_FILTER_VF;
+ __u32 filt_mask = 0;
struct nlmsghdr *answer;
+ if (!filter.vfinfo)
+ filt_mask |= RTEXT_FILTER_VF;
if (!show_stats)
filt_mask |= RTEXT_FILTER_SKIP_STATS;
-
addattr32(&req.n, sizeof(req), IFLA_EXT_MASK, filt_mask);
if (rtnl_talk(&rth, &req.n, &answer) < 0) {
@@ -2139,6 +2141,7 @@ static int ipaddr_list_flush_or_save(int argc, char **argv, int action)
ipaddr_reset_filter(oneline, 0);
filter.showqueue = 1;
filter.family = preferred_family;
+ filter.vfinfo = 0;
if (action == IPADD_FLUSH) {
if (argc <= 0) {
@@ -2221,6 +2224,8 @@ static int ipaddr_list_flush_or_save(int argc, char **argv, int action)
invarg("\"proto\" value is invalid\n", *argv);
filter.have_proto = true;
filter.proto = proto;
+ } else if (strcmp(*argv, "novf") == 0) {
+ filter.vfinfo = -1;
} else {
if (strcmp(*argv, "dev") == 0)
NEXT_ARG();
@@ -2274,7 +2279,7 @@ static int ipaddr_list_flush_or_save(int argc, char **argv, int action)
* the link device
*/
if (filter_dev && filter.group == -1 && do_link == 1) {
- if (iplink_get(filter_dev, RTEXT_FILTER_VF) < 0) {
+ if (iplink_get(filter_dev, filter.vfinfo < 0 ? 0 : RTEXT_FILTER_VF) < 0) {
perror("Cannot send link get request");
delete_json_obj();
exit(1);
diff --git a/ip/iplink.c b/ip/iplink.c
index 95314af5..1bb4a998 100644
--- a/ip/iplink.c
+++ b/ip/iplink.c
@@ -111,7 +111,7 @@ void iplink_usage(void)
" [ gro_max_size BYTES ] [ gro_ipv4_max_size BYTES ]\n"
"\n"
" ip link show [ DEVICE | group GROUP ] [up] [master DEV] [vrf NAME] [type TYPE]\n"
- " [nomaster]\n"
+ " [nomaster] [ novf ]\n"
"\n"
" ip link xstats type TYPE [ ARGS ]\n"
"\n"
diff --git a/man/man8/ip-link.8.in b/man/man8/ip-link.8.in
index 31e2d7f0..066ad874 100644
--- a/man/man8/ip-link.8.in
+++ b/man/man8/ip-link.8.in
@@ -202,7 +202,8 @@ ip-link \- network device configuration
.IR ETYPE " ] ["
.B vrf
.IR NAME " ] ["
-.BR nomaster " ]"
+.BR nomaster " ] ["
+.BR novf " ]"
.ti -8
.B ip link xstats
@@ -2898,6 +2899,10 @@ output.
.B nomaster
only show devices with no master
+.TP
+.B novf
+only show devices with no VF info
+
.SS ip link xstats - display extended statistics
.TP
--
2.33.0
Powered by blists - more mailing lists