Re: hot path optimizations in uma_zalloc() & uma_zfree()
Lately Nikos Ntarmos <[EMAIL PROTECTED]> said: > Ummm... What compiler version are you use? The pointer variant produces > faster code for me on both NetBSD/alpha (gcc 3.3.3) and Linux/x86 (gcc > 3.3.5), using both -O0 and -O2 (other compiler flags also tested with > similar results). Perhaps something specific to FreeBSD then? I guess > I'll have to set up a FBSD box and have a look at it, when I catch up > with some of the other stuff in my todo list... This is micro-optimization at its best. I doubt it has any influence in real world scenarios. And if it does, something else is extremely broken. You don't want to takt the route and implement it in assembler? Besides, I guess it's totally CPU dependant, how stuff can be pipelined, all memory locations ought to be in the cache anyways. cheers simon -- Serve - BSD +++ RENT this banner advert +++ASCII Ribbon /"\ Work - Mac +++ space for low $$$ NOW!1 +++ Campaign \ / Party Enjoy Relax | http://dragonflybsd.org Against HTML \ Dude 2c 2 the max ! http://golden-apple.biz Mail + News / \ pgpau0yQIT3TE.pgp Description: PGP signature
AT-1500FT PnP card, FreeBSD-5.4R-p1 and lnc(4) problems
Hi,
I am trying to make a network card work with lnc(4) device.
The card is an AT-1500 PnP Ethernet Adapter based on PCnet(TM)-ISA II
chip from AMD (AM79C961AKC) which is supposed to be supported under lnc(4)
driver. The card has both a 10BaseFL(ST) and 10BaseT port.
After adding the card's pnp id to the lnc_pnp_ids[] structure I am getting
the card recognized and attached but the chip and MAC address are wrongly
detected.
Both the Ethernet Address and ic type are wrongly detected!
Any help will be appreciated!
#uname -a
FreeBSD test 5.4-RELEASE-p1 FreeBSD 5.4-RELEASE-p1 #4: Sun Jun 5
00:18:56 CEST 2005 [EMAIL PROTECTED]:/usr/src/sys/i386/compile/TEST i386
#dmesg
...
lnc1: at port 0x220-0x237 irq 9 drq 3 on
isa0
lnc1: Attaching ATI AT-1500 Ethernet Network Adapter
lnc1: Ethernet address: cf:bf:ff:ef:00:ff <--wrong (tested against DOS setup
program)
lnc1: if_start running deferred for Giant
lnc1: Unknown <- (chip id 0!)
...
A diff to if_lnc_isa:
=
--- sys/dev/lnc/if_lnc_isa.c.orig Sun Jan 30 01:59:53 2005
+++ sys/dev/lnc/if_lnc_isa.cSun Jun 5 00:16:19 2005
@@ -54,6 +54,8 @@
#include
static struct isa_pnp_id lnc_pnp_ids[] = {
+{0x00158b06, NULL}, /* ATK1500 - ATI AT-1500 Ethernet Network
Adapter */
+{0x8c82d041, "AMD PCNet Family Cards"}, /* PNP828C */
{0, NULL}
};
=
the information added to if_lnc_isa.c was obtained with pnpinfo(8).
Regards,
G. Ardelean
___
freebsd-hackers@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-hackers
To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: Panic in bus_dmamap_load_mbuf (pmap.h:201)
On Sun, 3 Jul 2005 22:56:33 +0200 Cédric Jonas <[EMAIL PROTECTED]> wrote: > Hello, > > Since 1 or 2 months, I experienced panics when I use the ath interface of our > server (configured as hostap). Also, the system hangs often (without panic) > during the boot process at the "Setting hostname: icarus." level. Because I > had absolutely no time, I couldn't learn to use the kernel debugger before. > > > > icarus# uname -a > FreeBSD icarus 5.4-STABLE FreeBSD 5.4-STABLE #0: Sun Jul 3 15:14:13 > CEST 2005 [EMAIL PROTECTED]:/usr/obj/usr/src/sys/ICARUS i386 > > Sources are from July, 1 > > > debug.mpsafenet: 1 > > > > After compiling the kernel with debug options, I get the following during the > boot process: > > Setting hostname: icarus. > lock order reversal >1st 0xc2cd4c6c ath0 (network driver) @ > /usr/src/sys/dev/ath/if_ath.c:893 >2nd 0xc30583c8 user map (user map) @ /usr/src/sys/vm/vm_map.c:2998 > > > Another important detail could be that I use a dual Xeon box, which HT > enabled, which results in 4 logical CPU's and explain why anybody reported > this problem before?! > > I hope somebody could help, if you need additional informations, let me know. > Thank you. > > -- > Best regards, > Cédric Jonas > ___ > freebsd-hackers@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-hackers > To unsubscribe, send any mail to "[EMAIL PROTECTED]" > If it can help: I tried to disable PREEMPTION, the hyperthreading functionality (=> 2 logical CPU), and I set debug.mpsafenet to 0, but the LOR (+ Panic when ath0 is used) continue to occure... I tried also to downgrade to 5.3-RELEASE, where the problems do _not_ occur... but that's not a solution. -- Best regards, Cédric Jonas ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: buildworld sequence
Hi, > I looking for an explanation of the buildworld sequence : > legacy > bootstrap tools > build tools > cross tools > ... > > In fact, I read Makefile.inc1, so I have a basic idea of what each step > does, but I don't really understand > * what is the legacy step, > * why strfile(1), gperf(1) and so on are required to be built before > mostly everything else (bootstrap tools) > * why utilities in the build-tools target are built before gcc. > > I didn't asked all the questions I have in mind, but I would be glad if > someone could give me the general idea behind this or, even better, > point me some documentation. I checked developpers-handbook, there is > nothing about it, unfortunately. I overlook the description of each stage at the beginning of the file, I'll whip myself for this. > I was trying to understand this because I need to use a gcc(1) option > in CFLAGS for these tools for my integration of the ProPolice patch > in the build system. I tried to set CFLAGS as an environnement variable in BMAKEENV from Makefile.inc1, but the compilation line still doesn't't contain the flag I want. Any idea why this doesn't work ? Regards, -- Jeremie Le Hen < jeremie at le-hen dot org >< ttz at chchile dot org > ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
