netdev - Re: [PATCH net-next,v2] IPv6 ifstats separation

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <cdf02f85-57ea-3371-cec3-ec155964f4d0@gmail.com>
Date:   Thu, 4 Oct 2018 13:42:06 -0700
From:   Eric Dumazet <eric.dumazet@...il.com>
To:     Stephen Suryaputra <ssuryaextr@...il.com>, netdev@...r.kernel.org
Subject: Re: [PATCH net-next,v2] IPv6 ifstats separation



On 10/04/2018 12:08 PM, Stephen Suryaputra wrote:
> Separate IPv6 ifstats into the ones that are hit on fast path and
> the ones that aren't. The ones that are not can be removed as needed
> using sysctls.
> 
> The separation minimizes percpu data. It reduces from 288 bytes per cpu to 56
> bytes. When the other stats are disabled, then it adds 4144 bytes saving per
> netdev.
> 
> The stats that are hit on fast path are percpu and the existing macro that
> increments them is added with a constant check that should be optimized by the
> preprocessor.
> 
> This is done based on some feedbacks in the past to make IPv6 stats optional.
> It doesn't fully accomplish that but it enables all but 6 counters be optional.
> Those optional counters however are still enabled by default to preserve the
> current behavior.
> 

How have you decided some counters can be 'slow' and other 'fast' ?

I can tell you I see many ultra-fast candidates in your 'slow' list :/

Also think about DDOS.

After your patch, all these 'wrong packets' will incur an expensive
operation on a shared and highly contented cache line,
effectively making the attack easier to conduct.


+static const struct snmp_mib snmp6_ipstats_device_list[] = {
+	SNMP_MIB_ITEM("Ip6InHdrErrors", IPSTATS_MIB_INHDRERRORS),
+	SNMP_MIB_ITEM("Ip6InTooBigErrors", IPSTATS_MIB_INTOOBIGERRORS),
+	SNMP_MIB_ITEM("Ip6InNoRoutes", IPSTATS_MIB_INNOROUTES),
+	SNMP_MIB_ITEM("Ip6InAddrErrors", IPSTATS_MIB_INADDRERRORS),
+	SNMP_MIB_ITEM("Ip6InUnknownProtos", IPSTATS_MIB_INUNKNOWNPROTOS),
+	SNMP_MIB_ITEM("Ip6InTruncatedPkts", IPSTATS_MIB_INTRUNCATEDPKTS),
+	SNMP_MIB_ITEM("Ip6InDiscards", IPSTATS_MIB_INDISCARDS),
+	SNMP_MIB_ITEM("Ip6OutDiscards", IPSTATS_MIB_OUTDISCARDS),
+	SNMP_MIB_ITEM("Ip6OutNoRoutes", IPSTATS_MIB_OUTNOROUTES),
+	SNMP_MIB_ITEM("Ip6ReasmTimeout", IPSTATS_MIB_REASMTIMEOUT),
+	SNMP_MIB_ITEM("Ip6ReasmReqds", IPSTATS_MIB_REASMREQDS),
+	SNMP_MIB_ITEM("Ip6ReasmOKs", IPSTATS_MIB_REASMOKS),
+	SNMP_MIB_ITEM("Ip6ReasmFails", IPSTATS_MIB_REASMFAILS),
+	SNMP_MIB_ITEM("Ip6FragOKs", IPSTATS_MIB_FRAGOKS),
+	SNMP_MIB_ITEM("Ip6FragFails", IPSTATS_MIB_FRAGFAILS),
+	SNMP_MIB_ITEM("Ip6FragCreates", IPSTATS_MIB_FRAGCREATES),
+	SNMP_MIB_ITEM("Ip6InMcastPkts", IPSTATS_MIB_INMCASTPKTS),
+	SNMP_MIB_ITEM("Ip6OutMcastPkts", IPSTATS_MIB_OUTMCASTPKTS),
+	SNMP_MIB_ITEM("Ip6InMcastOctets", IPSTATS_MIB_INMCASTOCTETS),
+	SNMP_MIB_ITEM("Ip6OutMcastOctets", IPSTATS_MIB_OUTMCASTOCTETS),
+	SNMP_MIB_ITEM("Ip6InBcastOctets", IPSTATS_MIB_INBCASTOCTETS),
+	SNMP_MIB_ITEM("Ip6OutBcastOctets", IPSTATS_MIB_OUTBCASTOCTETS),
+	/* IPSTATS_MIB_CSUMERRORS is not relevant in IPv6 (no checksum) */
+	SNMP_MIB_ITEM("Ip6InNoECTPkts", IPSTATS_MIB_NOECTPKTS),
+	SNMP_MIB_ITEM("Ip6InECT1Pkts", IPSTATS_MIB_ECT1PKTS),
+	SNMP_MIB_ITEM("Ip6InECT0Pkts", IPSTATS_MIB_ECT0PKTS),
+	SNMP_MIB_ITEM("Ip6InCEPkts", IPSTATS_MIB_CEPKTS),
+	SNMP_MIB_SENTINEL
+};
+