(Logical change 1.3)

[mirror_iproute2.git] / Patches / Catalogue

diff --git a/Patches/Catalogue b/Patches/Catalogue
index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..8e1927914b5701ec46d6f2152225e46f0d4662ec 100644 (file)
--- a/Patches/Catalogue
+++ b/Patches/Catalogue
@@ -0,0 +1,46 @@
+File:          rt_cache_stat.dif
+Apply to:      kernel < 2.4.7
+Status:                recommended for kernels < 2.4.7.
+               already present in >= 2.4.7
+Description:   tracing efficiency of routing cache
+Side effects:  none
+
+File:          pidentd-3.0.12.dif
+Apply to:      pident-3.0.12 tree f.e. from am redhat rpm
+Status:                highly recommended
+Description:   Patch to pidentd allowing to use tcpdiag facility and fixing
+               some bugs in original pident.
+Side effects:  none. Does not break anything not depending on kernel version,
+               even if tcpdiag is absent.
+Advice:                not related to this patch but should be said yet.
+               Do NOT configure pidentd to use threads! Use option
+               "--without-threads" when doing "configure".
+               pidentd is typical example of application where
+               threading results in nothing but collapse of performance.
+               Apparently author learned thread programming and decided
+               to apply new knowledge to the first victim.
+
+File:          symbol_exports.dif 
+Apply to:      kernel < 2.4.17
+Status:                desired for kernels < 2.4.17
+               not needed for kernels >= 2.4.17
+Description:   exports symbols required to load tcpdiag module
+               tcpdiag is builtin since 2.4.17, hence the exports
+               are redundant.
+Side effects:  none
+
+File:          af_unix.dif
+Apply to:      kernel
+Status:                recommended
+Desciption:    implements fragmented skb for unix sockets reducing
+               vm pressure for datagram sockets and adds to /proc/net/unix
+               columns allowing to monitor recv/send memory and identify
+               peer of connected sockets.
+Side effects:  "lsof" blames something about unix sockets.
+               Not a big loss, lsof is not able to tell anything more
+               clever than "can't identify protocol" for sockets anyway. 
+Note:          the patch affects area where one or two lines changed
+               several times while 2.4. It does not depend on this,
+               but unfortunately may reject. It apply cleanly to
+               2.4.17.
+