| To: | jamal <hadi@xxxxxxxxxx> |
|---|---|
| Subject: | Re: [PATCH] improvement on net/sched/cls_fw.c's hash function |
| From: | Thomas Graf <tgraf@xxxxxxx> |
| Date: | Wed, 6 Apr 2005 16:10:20 +0200 |
| Cc: | Wang Jian <lark@xxxxxxxxxxxx>, netdev <netdev@xxxxxxxxxxx> |
| In-reply-to: | <20050406134502.GP26731@xxxxxxxxxxxxxx> |
| References: | <20050405213023.0256.LARK@xxxxxxxxxxxx> <1112717495.1076.22.camel@xxxxxxxxxxxxxxxx> <20050406143842.026B.LARK@xxxxxxxxxxxx> <20050406123036.GO26731@xxxxxxxxxxxxxx> <1112794459.1096.61.camel@xxxxxxxxxxxxxxxx> <20050406134502.GP26731@xxxxxxxxxxxxxx> |
| Sender: | netdev-bounce@xxxxxxxxxxx |
* Thomas Graf <20050406134502.GP26731@xxxxxxxxxxxxxx> 2005-04-06 15:45 > * jamal <1112794459.1096.61.camel@xxxxxxxxxxxxxxxx> 2005-04-06 09:34 > > When does the user specify this? i would think you need to set it once > > only per bootup. After that it will be quiet complex to reset. It could > > be done but sounds complex and unnecessary. i.e it may be a boot or > > module parameter more than it is a netlink controlled value, no? > > I'm not 100% sure about this yet but I think during fw_change so > we can have different hashes for different qdiscs filter chains. The thing I'm worrying about is that I don't want to break the perfect alignment of fw_head to good slab obj sizes but I guess there is no way around. I'd really like to make hash size and hash function configureable. For example a hash size of 1024 would perform much better and would still fit into a single page on most systems. |
| Previous by Date: | Re: [PATCH] improvement on net/sched/cls_fw.c's hash function, Wang Jian |
|---|---|
| Next by Date: | Re: e1000 packet corruption problem, Michal Vanco |
| Previous by Thread: | Re: [PATCH] improvement on net/sched/cls_fw.c's hash function, Thomas Graf |
| Next by Thread: | Re: [PATCH] improvement on net/sched/cls_fw.c's hash function, David S. Miller |
| Indexes: | [Date] [Thread] [Top] [All Lists] |