Re: git: 9b8db66402e9 - main - netgraph: provide separate malloc type for nodes that are missing it

<div><div>I think we should remove this option altogether - NG_SEPARATE_MALLOC.</div><div> </div><div>Now, 90+% of nodes declare their own malloc(9) type.</div><div> </div><div>For example, look at https://github.com/freebsd/freebsd-src/blob/main/sys/netgraph/ng_bridge.c#L81</div><div> </div><div>Some people want to see this through vmstat.</div></div><div> </div><div> </div><div>31.05.2024, 21:03, "Brooks Davis" &lt;brooks@freebsd.org&gt;:</div><div><br />On Fri, May 31, 2024 at 04:20:02PM +0000, Gleb Smirnoff wrote:<br /> The branch main has been updated by glebius:<br /> <br /> URL: https://cgit.FreeBSD.org/src/commit/?id=9b8db66402e94919ab47fc2d69863968e49e305a<br /> <br /> commit 9b8db66402e94919ab47fc2d69863968e49e305a<br /> Author: Dmitry Lukhtionov &lt;dmitryluhtionov@gmail.com&gt;<br /> AuthorDate: 2024-05-31 16:19:54 +0000<br /> Commit: Gleb Smirnoff &lt;glebius@FreeBSD.org&gt;<br /> CommitDate: 2024-05-31 16:19:54 +0000<br /> <br />     netgraph: provide separate malloc type for nodes that are missing it<br />     <br />     The kernel option NG_SEPARATE_MALLOC helps to debug memory leaks in<br />     netgraph(4). Several nodes were missing the support.<br /><br />I realize this commit didn't create the option, but is there really<br />enough of a cost to having multiple malloc types to justify an option at<br />all?<br /><br />-- Brooks</div>