| To: | Ben Greear <greearb@xxxxxxxxxxxxxxx> |
|---|---|
| Subject: | Re: [PATCH] pktgen: reduce stack usage |
| From: | "Randy.Dunlap" <rddunlap@xxxxxxxx> |
| Date: | Mon, 21 Feb 2005 08:27:20 -0800 |
| Cc: | robert.olsson@xxxxxxxxx, netdev <netdev@xxxxxxxxxxx> |
| In-reply-to: | <4216AAE4.1080804@xxxxxxxxxxxxxxx> |
| Organization: | OSDL |
| References: | <20050218134219.3f079110.rddunlap@xxxxxxxx> <4216AAE4.1080804@xxxxxxxxxxxxxxx> |
| Sender: | netdev-bounce@xxxxxxxxxxx |
| User-agent: | Mozilla Thunderbird 1.0 (X11/20041206) |
Ben Greear wrote: Randy.Dunlap wrote:(resend) Any comments this time? pktgen: proc_if_write: reduce stack usage (on i386) from 776 to 296 bytes by combining/reusing locals.Since these methods are not in the fast path, and the stack usage is not near 4k, does this really matter?
It's not critical or near causing a stack overflow AFAICT.
And ideally gcc would recognize this kind of usage and not
allocate multiple stack entries for it.
syscall: sys_write (tiny stack usage)
-> vfs_write (tiny)
-> proc_file_write (tiny)
-> proc_if_write (pktgen: large)
I'm asking more to be educated than to dismiss the changes.... -- ~Randy |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: [PATCH] pktgen: reduce stack usage, Randy.Dunlap |
|---|---|
| Next by Date: | Re: Intel and TOE in the news, Thomas Graf |
| Previous by Thread: | Re: [PATCH] pktgen: reduce stack usage, Ben Greear |
| Next by Thread: | Intel and TOE in the news, Jeff Garzik |
| Indexes: | [Date] [Thread] [Top] [All Lists] |