| TJ- | What's the best way to effectively disable IPv4 as comprehensively as possible given there's no generic CONFIG_IPV4=n or /proc/sys/net/ipv4/conf/all/disable_ipv4 ? | 06:36 |
|---|---|---|
| cascardo | iptables comes to mind | 13:00 |
| cascardo | enyc: look at /usr/share/doc/linux-modules-5.4.0-53-generic/changelog.Debian.gz | 13:00 |
| enyc | cascardo: OOOOOoOOOOOOO thankyou for pointer! | 13:08 |
| LordDoskias | is ubuntu 20l.04 supposed to support amd ibs ? support for it landed many years ago | 15:47 |
| LordDoskias | but still when i run: perf record -e cpu-cycles i get an EINVAL error when passing the :p modifier ? | 15:48 |
| LordDoskias | i also get the following in dmesg: [чт ное 12 17:11:52 2020] perf: AMD IBS detected (0x000003ff) | 15:49 |
| LordDoskias | 15:49 | |
| LordDoskias | so it seems support should be there yet it doesn't work | 15:49 |
| cascardo | LordDoskias: try it with --count=1000 | 18:18 |
| LordDoskias | sudo perf record -c 1000 -g -e cpu-cycles:pp doesn't work | 18:19 |
| LordDoskias | but this one does: sudo perf record -c 100001 -e ibs_op/cnt_ctl=1/pp -a -C 0 taskset -c 0 | 18:19 |
| LordDoskias | ... | 18:19 |
| LordDoskias | amd'\s perf guy gave me this command line, apparently IBS is b0rked ... | 18:24 |
| cascardo | LordDoskias: what about with -F and -c ? looking at the IBS code here | 18:33 |
| LordDoskias | with just -c1000 still doesn't work | 18:36 |
| LordDoskias | i think the even is different | 18:36 |
| LordDoskias | i.e cpu-cycle:pp is not doing what's supposed | 18:37 |
| LordDoskias | only -e ibs_op/cntl_ctl=1 does ... | 18:37 |
| cascardo | what about -e r076:p ? | 18:50 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!