Re: Balancing Outgoing traffic over 2 nics, and nic limitations.
> This configuration is neccessary because by my estimation I have run > into a limit on the intel pro 100 netcards of 6,000 packets/second. > This limit equates to about 30 to 32 megabit/second of web traffic in > our situation. I am wondering if anyone else has noticed this limit? The Pro 100B/Pro 100+ models, using the fxp driver, are fully capable of saturating a 100 Mbps Ethernet. With maximum sized packets, this gives you more than 8000 packets per second. I have measured this many times myself. Thus I think the limitation you're seeing is not in the Intel Ethernet cards. Steinar Haug, Nethelp consulting, [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: SUIDDIR problem
> SUIDDIR will work for any user EXCEPT ROOT > I did this because I felt it was a security hole to allow users to create > files owned by root. > (from memory it will also refuse to do files that have the execute bit set > but I can't remember for sure) In a mildly drunken state, I respond. :) Without looking, I'd imagine that if the chmod command of FTP will allow you to do a "chmod 4755 file-I-just-uploaded" -- if you have the ability to execute programs on the machine you uploaded to, this could be a major problem. Hence, I'd agree with your decision. --mike To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: SUIDDIR problem
On Sat, 16 Oct 1999, Mike Nowlin wrote: > > > SUIDDIR will work for any user EXCEPT ROOT > > I did this because I felt it was a security hole to allow users to create > > files owned by root. > > (from memory it will also refuse to do files that have the execute bit set > > but I can't remember for sure) > > In a mildly drunken state, I respond. :) > > Without looking, I'd imagine that if the chmod command of FTP will allow > you to do a "chmod 4755 file-I-just-uploaded" -- if you have the ability > to execute programs on the machine you uploaded to, this could be a major > problem. Hence, I'd agree with your decision. Since the ftp daemon will create files without the x bits set, they will succeeed, and will immediatly be owned by the owner of the directory. The sender no-longer owns them and cannot set mode bits, whether or not the ftp daemon would allow it to. > > --mike > > > To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: FreeBSD and HP Jornado
On Fri, 15 Oct 1999 17:29:07 -0700 Edward Elhauge <[EMAIL PROTECTED]> wrote: > My other questions is if is a way of replacing the CE OS with something > easier to customize and that might run either Perl or Java? What sort of processor does the Jornado have? If it's a MIPS-based machine, getting it to run NetBSD/hpcmips might be a possibility. -- Jason R. Thorpe <[EMAIL PROTECTED]> To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
ftpmirror proxy
I realy like ftpmirror it is a great program. I was wondering if ftpmirror can use a squid proxy server which runs on port 3128? Is there a way to use ftp-gateway on a specific port? below is a sampe of the configurations I have tried. Thank you for your time Richard Puga [EMAIL PROTECTED] ftp-user = anonymous ftp-pass = [EMAIL PROTECTED] ftp-stats = yes temp-directory = /usr/tmp package = FreeBSD ftp-server = ftp.freebsd.org # ftp-gateway = ftp.squid.proxy.server 3128 ftp-gateway = ftp.squid.proxy.server :3128 remote-directory = /pub/FreeBSD local-directory = /usr/var/ftp/pub/FreeBSD transfer-file-regexp += !/\/core$/ transfer-file-regexp += !/\.core$/ transfer-file-regexp += !/\/\ko-/ transfer-file-regexp += !/\/\ja-/ transfer-file-regexp += !/\/\zh-/ transfer-directory-regexp += !/\/CERT\/$/ transfer-directory-regexp += !/\/releases\/alpha\/$/ transfer-directory-regexp += !/\/development\/$/ transfer-directory-regexp += !/\/doc\/mailing\-lists\/$/ transfer-directory-regexp += !/\/ports\/alpha\/$/ transfer-directory-regexp += !/\/ports\/i386\/packages\-3\.0\-aout\/$/ transfer-directory-regexp += !/\/branches\/\-current\/XF86333\/$/ transfer-directory-regexp += !/\/branches\/\-current\/packages\-aout\/ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: FreeBSD and HP Jornado
Jason R. Thorpe <[EMAIL PROTECTED]> wrote: > Edward Elhauge <[EMAIL PROTECTED]> wrote: >> My other questions is if is a way of replacing the CE OS with something >> easier to customize and that might run either Perl or Java? >What sort of processor does the Jornado have? If it's a MIPS-based machine, >getting it to run NetBSD/hpcmips might be a possibility. The high end one (820 ?) has a 190MHz SA1100 StrongArm. I don't think that there is any support in NetBSD/arm32 for either the SA1100 or SA1110. Robert Swindells To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: Balancing Outgoing traffic over 2 nics, and nic limitations.
> Hello my FreeBSD friends. > > I have two issues. > > The first is how to balance outbound traffic over 2 nics that are on > the same subnet. Example configuration: > > fxp0: 12.2.2.5 netmask 255.255.255.0 > fxp1: 12.2.2.6 netmask 255.255.255.255 > > router at: 12.2.2.1 You can't do this. If all of the outbound traffic is headed for the same router, put two cards in the router and use two separate nets. > Currently I have the obvious static route to 12.2.2.1, which locks onto > fxp0 so all outbound traffic flows out over that link. Inbound traffic > balances per ip as I would expect. I hope to find a scalable solution > as I hope to build a server that will utilize 3 nics. This is not a sensible course of action. > This configuration is neccessary because by my estimation I have run > into a limit on the intel pro 100 netcards of 6,000 packets/second. These cards do not exhibit such a limit. You may have run into some issues with FreeBSD's ability to handle very large numbers of small packets with your particular application mix. > This limit equates to about 30 to 32 megabit/second of web traffic in > our situation. I am wondering if anyone else has noticed this limit? Not in my recollection. The fxp driver in recent incarnations limits the number of interrupts it generates by restricting them to low-resource conditions rather than generating one per packet. And I've personally seen an SMP kernel run tolerably while taking > 100,000 interrupts per second. > This limit was hit on 2 very different machines, one with significantly > less power. Any feedback on either of these issues would be > appreciated. I'd start by eliminating the network adapter and driver; move to an up-to-date FreeBSD-stable and substitute a 3C905B or C and determine for yourselves whether this is really an issue with the card. General experience would suggest that you should be able to come close to saturating your network with even relatively small datagrams using either of these adapters. You also don't mention whether you're running on a switched network; at that sort of traffic level you will definitely want to be using a switch that supports full duplex operation. -- \\ Give a man a fish, and you feed him for a day. \\ Mike Smith \\ Tell him he should learn how to fish himself, \\ [EMAIL PROTECTED] \\ and he'll hate you for a lifetime. \\ [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: how mkdir without .. ?
"Aleksandr A.Babaylov" wrote: > > I need in directories without link to parent in it > or with link to parent renamed to something exotic name. What are you trying to accomplish? If you are trying to create directories that users cannot "escape" out of, all you need to do is remove the world rx permissions on the parent directory, and make sure that they don't have access via the group permissions. You might also want to look at chroot. Good luck, Doug -- "Stop it, I'm gettin' misty." - Mel Gibson as Porter, "Payback" To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
I can't boot FreeBSD 3.2 from Installation CD (fwd
Hi, Sorry, I have sent this e-mail to questions from monday but no I haven't received any answer. Thanks for your help. Original Message Subject: I can't boot FreeBSD 3.2 from Installation CD (fwd) Date: Mon, 11 Oct 1999 16:25:07 -0500 (COT) From: Yonny Cardenas Baron <[EMAIL PROTECTED]> To: [EMAIL PROTECTED] I have a problem with the installation of FreeBSD Releases 3.1 and 3.2. My box can't boot from CD or installation floppies, the following message was showed: -- BTX loader BIOS drive A: is disk0 BIOS drive B: is disk1 Can't work out which disk we are booting from Guessed BIOS device 0x8b not found by probes defaulting to disk0: - Now, I is running well FreeBSD 2.2.5 this computer, I can boot from CD and floppies installation 2.2.5. It is a IBM PC Personal Computer 300GL with Pentium II, RAM 64M and HD 3Gb. Thanks for your help. YONNY CARDENAS B. Escuela Colombiana de Ingenieria e-mail: [EMAIL PROTECTED] [EMAIL PROTECTED] = __ Do You Yahoo!? Bid and sell for free at http://auctions.yahoo.com To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
MAXPATHLEN not enforced
Why does FreeBSD let you create paths longer than MAXPATHLEN?
I often have various trees that are as deep as possible for testing various
programs for holes, and I finally figured out why locate wasn't updating its
database properly; it was choking as soon as it saw a path length
>MAXPATHLEN long. The question, however, is why can it see a path length
longer than MAXPATHLEN?
I would also wonder if there aren't some security issues resulting
from this. From what gdb shows, locate seems to trash its stack
before spitting out the error about the path being too long...
marcs@alive:/tmp$ mkdir erm...
marcs@alive:/tmp$ cd erm...
marcs@alive:/tmp/erm...$ while mkdir x; do cd x; done
cd: could not get current directory: getcwd: cannot access parent directories: Result
too large
cd: could not get current directory: getcwd: cannot access parent directories: Result
too large
^Cjob-working-directory: could not get current directory: getcwd: cannot access parent
directories: Result too large
cd_links: could not get current directory: getcwd: cannot access parent directories:
Result too large
^C^C^C^C^Ccd: could not get current directory: getcwd: cannot access parent
directories: Result too large
^C^C^C^C^Ccd: could not get current directory: getcwd: cannot access parent
directories: Result too large
^C^C^C^C^C^C^C^C^C^C^C^C^C
^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C
marcs@alive:.$ cd /
cd_links: could not get current directory: getcwd: cannot access parent directories:
Result too large
marcs@alive:/$ find /tmp/erm.../ | perl -ne 'if (length($_) > 1024) { print
length($_), ": $_\n" }'
(...a few other results...)
1038:
/tmp/erm.../x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x/x
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
Re: Balancing Outgoing traffic over 2 nics, and nic limitations.
On 10/16/99 at 10:35 AM Mike Smith wrote: >> Hello my FreeBSD friends. >> >> I have two issues. >> >> The first is how to balance outbound traffic over 2 nics that are on >> the same subnet. Example configuration: >> >> fxp0: 12.2.2.5 netmask 255.255.255.0 >> fxp1: 12.2.2.6 netmask 255.255.255.255 >> >> router at: 12.2.2.1 > > >You can't do this. If all of the outbound traffic is headed for the >same router, put two cards in the router and use two separate nets. > >> Currently I have the obvious static route to 12.2.2.1, which locks onto >> fxp0 so all outbound traffic flows out over that link. Inbound traffic >> balances per ip as I would expect. I hope to find a scalable solution >> as I hope to build a server that will utilize 3 nics. > >This is not a sensible course of action. > Sounds fair. :) >> This configuration is neccessary because by my estimation I have run >> into a limit on the intel pro 100 netcards of 6,000 packets/second. > >These cards do not exhibit such a limit. You may have run into some >issues with FreeBSD's ability to handle very large numbers of small >packets with your particular application mix. > >> This limit equates to about 30 to 32 megabit/second of web traffic in >> our situation. I am wondering if anyone else has noticed this limit? > >Not in my recollection. The fxp driver in recent incarnations limits >the number of interrupts it generates by restricting them to >low-resource conditions rather than generating one per packet. And >I've personally seen an SMP kernel run tolerably while taking > 100,000 >interrupts per second. > >> This limit was hit on 2 very different machines, one with significantly >> less power. Any feedback on either of these issues would be >> appreciated. > >I'd start by eliminating the network adapter and driver; move to an >up-to-date FreeBSD-stable and substitute a 3C905B or C and determine >for yourselves whether this is really an issue with the card. > >General experience would suggest that you should be able to come close >to saturating your network with even relatively small datagrams using >either of these adapters. > >You also don't mention whether you're running on a switched network; at >that sort of traffic level you will definitely want to be using a >switch that supports full duplex operation. Of course its a switched network with full duplex operation. But now that the general answer is that it is not a limitation of the nic card I am going to look elsewhere. I was not to sure if it was actually a limit myself, its just that I observed it on two different machines. They however were not huge powerhouses, one was a p2-450, and one was a dual p2 333. Both running real new versions of 3.3-stable. Thanks for the replies at any rate. To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: FreeBSD and HP Jornado
On Sat, 16 Oct 1999 17:57:37 +0100 (BST) Robert Swindells <[EMAIL PROTECTED]> wrote: > The high end one (820 ?) has a 190MHz SA1100 StrongArm. > > I don't think that there is any support in NetBSD/arm32 for either the > SA1100 or SA1110. No, but it probably wouldn't be that hard to make it go :-) -- Jason R. Thorpe <[EMAIL PROTECTED]> To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: Balancing Outgoing traffic over 2 nics, and nic limitations.
> > Of course its a switched network with full duplex operation. But now > that the general answer is that it is not a limitation of the nic card > I am going to look elsewhere. I was not to sure if it was actually a > limit myself, its just that I observed it on two different machines. You should be looking at some basic system statistics to see where your limiting factor actually is; try watching 'systat -vmstat 1' while you're hitting your limit. > They however were not huge powerhouses, one was a p2-450, and one was a > dual p2 333. Both running real new versions of 3.3-stable. Note also that Apache isn't the fastest of animals; you might want to use a lighter-weight server like thttpd if all you're doing is serving static content. -- \\ Give a man a fish, and you feed him for a day. \\ Mike Smith \\ Tell him he should learn how to fish himself, \\ [EMAIL PROTECTED] \\ and he'll hate you for a lifetime. \\ [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: Balancing Outgoing traffic over 2 nics, and nic limitations.
> Of course its a switched network with full duplex operation. But now > that the general answer is that it is not a limitation of the nic card > I am going to look elsewhere. I was not to sure if it was actually a > limit myself, its just that I observed it on two different machines. > They however were not huge powerhouses, one was a p2-450, and one was a > dual p2 333. Both running real new versions of 3.3-stable. FWIW, FreeBSD 3.x with an Intel Pro 100B/100+ card can saturate a 100 Mbps Ethernet with something like a P-166. This is with maximum sized frames, running ttcp or Netperf. You *don't* need a huge powerhouse with FreeBSD :-) Steinar Haug, Nethelp consulting, [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: Balancing Outgoing traffic over 2 nics, and nic limitations.
> > Of course its a switched network with full duplex operation. But now > > that the general answer is that it is not a limitation of the nic card > > I am going to look elsewhere. I was not to sure if it was actually a > > limit myself, its just that I observed it on two different machines. > > They however were not huge powerhouses, one was a p2-450, and one was a > > dual p2 333. Both running real new versions of 3.3-stable. > > FWIW, FreeBSD 3.x with an Intel Pro 100B/100+ card can saturate a 100 Mbps > Ethernet with something like a P-166. This is with maximum sized frames, > running ttcp or Netperf. You *don't* need a huge powerhouse with FreeBSD > :-) The issue here isn't big frames though, it's little frames. You don't appear to have noticed that, and it's potentially very relevant. -- \\ Give a man a fish, and you feed him for a day. \\ Mike Smith \\ Tell him he should learn how to fish himself, \\ [EMAIL PROTECTED] \\ and he'll hate you for a lifetime. \\ [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: Status of UMAPFS
On Fri, 15 Oct 1999, Zhihui Zhang wrote: > > Is the UMAPFS working? I add "options UMAPFS" to the configuration file > of FreeBSD 3.3-Release and rebuilt the kernel. I got the following > errors: > > loading kernel > umap_vnops.o: In function `umap_lock': > umap_vnops.o(.text+0x568): undefined reference to `null_bypass' > umap_vnops.o: In function `umap_unlock': > umap_vnops.o(.text+0x58e): undefined reference to `null_bypass' > *** Error code 1 > > Stop. > I find out that you must also include NULLFS in the kernel to compile. I have tested NULLFS and UMAPFS with some trivial commands. Both works. -Zhihui To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: Balancing Outgoing traffic over 2 nics, and nic limitations.
On 10/16/99 at 11:26 AM Mike Smith wrote: >> >> Of course its a switched network with full duplex operation. But now >> that the general answer is that it is not a limitation of the nic card >> I am going to look elsewhere. I was not to sure if it was actually a >> limit myself, its just that I observed it on two different machines. > >You should be looking at some basic system statistics to see where your >limiting factor actually is; try watching 'systat -vmstat 1' while >you're hitting your limit. > >> They however were not huge powerhouses, one was a p2-450, and one was a >> dual p2 333. Both running real new versions of 3.3-stable. > >Note also that Apache isn't the fastest of animals; you might want to >use a lighter-weight server like thttpd if all you're doing is serving >static content. > We are actually using Zeus web server. Which is based on the select()/poll() model much like thttpd. Its extremely fast. It is also fully featured like apache, but 5 times faster. And boy, you are right about apache not being the fastest of animals. :) And to everyone following, this is web traffic, not huge tcp benchmark packets. Its not so bad now, we can push 35 megabit from one machine, where before apache would do about 10, but we could get it to 25 by putting a huge load on the machine. But I will look at systat now. Thanks To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: SMP + fxp0 wierdness
Mike Smith wrote: > > > Well that's not good, since I have almost convinced my boss to replace > > the crappy IDE drives on our shiny new Intel N440BX mb's with scsi > > drives since the controller is built in. :-/ Does this look like a > > soluble problem, or is it just going to be a case of "don't do that?" > > Anything I can do to help mail me and let me know. > > You could try turning on DMA and discovering that these drives aren't > _that_ shitty. 8) Heh... well I've already enabled flags 0xb0ff, which has improved things somewhat, but our hardware vendor slipped in some IBM DeskStar drives on us, and they've been no end of trouble. He _may_ live to regret his "mistake." The good news is that the first machine rebuilt with the SCSI drives and using ncr0 + fxp0 is not showing any signs of trouble. Crossing fingers, Doug To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: Balancing Outgoing traffic over 2 nics, and niclimitations.
Jimbo Bahooli wrote: > > Of course its a switched network with full duplex operation. But now > that the general answer is that it is not a limitation of the nic card > I am going to look elsewhere. I was not to sure if it was actually a > limit myself, its just that I observed it on two different machines. > They however were not huge powerhouses, one was a p2-450, and one was a > dual p2 333. Both running real new versions of 3.3-stable. No possibility of moving to a Gigabit adapter and sticking a Gigabit blade in your switch? Another possibility, depending on your switch vendor, may be to query them about some sort of channel aggregation. They might even be willing to open their source code -- who knows? Stranger things have happened. -- "Where am I, and what am I doing in this handbasket?" Wes Peters Softweyr LLC [EMAIL PROTECTED] http://softweyr.com/ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: FreeBSD and HP Jornado
Jason Thorpe wrote: > > On Sat, 16 Oct 1999 17:57:37 +0100 (BST) > Robert Swindells <[EMAIL PROTECTED]> wrote: > > > The high end one (820 ?) has a 190MHz SA1100 StrongArm. > > > > I don't think that there is any support in NetBSD/arm32 for either the > > SA1100 or SA1110. > > No, but it probably wouldn't be that hard to make it go :-) The Digital DNARD ran NetBSD/arm, but it was an SA-110. The 1100 is the same processor with more "system support" built in; it shouldn't be much of a leap. Jornada/BSD would be killer. -- "Where am I, and what am I doing in this handbasket?" Wes Peters Softweyr LLC [EMAIL PROTECTED] http://softweyr.com/ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: Handling segV's
Kevin Day wrote: > > I mmap() files in, then copy them to a device. This works great except when > someone tries to change that file during the copy. If the size of the file > shrinks, I'll SIGBUS or SIGSEGV when i try to touch past the new file size. > So, i setup a signal handler and longjmp into some recovery code. > > Perhaps there's a better way, but I don't consider this a bug really, and > i'll get a SEGV. :) MAP_PRIVATE? -- "Where am I, and what am I doing in this handbasket?" Wes Peters Softweyr LLC [EMAIL PROTECTED] http://softweyr.com/ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: Handling segV's
> > Kevin Day wrote: > > > > I mmap() files in, then copy them to a device. This works great except when > > someone tries to change that file during the copy. If the size of the file > > shrinks, I'll SIGBUS or SIGSEGV when i try to touch past the new file size. > > So, i setup a signal handler and longjmp into some recovery code. > > > > Perhaps there's a better way, but I don't consider this a bug really, and > > i'll get a SEGV. :) > > MAP_PRIVATE? > This has the undesired effect of me not noticing that the file changed then. Kevin To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: Search a symbol in the source tree
I wrote: > > Looking for where "utmp.h" is used: > > > > global -x -s utmp.h > > > > This takes more than 2212 seconds (over 36 minutes!), and outputs > > It seems that something wrong (bug?) occurred. > Would you please tell me the version of FreeBSD and GLOBAL? A structural problem was found in GLOBAL. I will optimize it. Thanks Darryl for his report. [work around] When you use regular expressions, please put prefix '^' to it like: global -x -s '^utmp.h' '.' is a regular expression which means any character. So, 'utmp.h' will match with 'aaautmpbhccc'. The '^' enables global to use index effectively. -- Shigio Yamaguchi - Tama Communications Corporation Mail: [EMAIL PROTECTED], WWW: http://www.tamacom.com To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: Handling segV's
Kevin Day wrote: > > > > > Kevin Day wrote: > > > > > > I mmap() files in, then copy them to a device. This works great except when > > > someone tries to change that file during the copy. If the size of the file > > > shrinks, I'll SIGBUS or SIGSEGV when i try to touch past the new file size. > > > So, i setup a signal handler and longjmp into some recovery code. > > > > > > Perhaps there's a better way, but I don't consider this a bug really, and > > > i'll get a SEGV. :) > > > > MAP_PRIVATE? > > > > This has the undesired effect of me not noticing that the file changed then. stat(2) it after the copy? OK, I'm reaching. Handling SEGV seems like a rather iffy solution at best. -- "Where am I, and what am I doing in this handbasket?" Wes Peters Softweyr LLC [EMAIL PROTECTED] http://softweyr.com/ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: Handling segV's
> > > > I mmap() files in, then copy them to a device. This works great except when > > > > someone tries to change that file during the copy. If the size of the file > > > > shrinks, I'll SIGBUS or SIGSEGV when i try to touch past the new file size. > > > > So, i setup a signal handler and longjmp into some recovery code. > > > > > > > > Perhaps there's a better way, but I don't consider this a bug really, and > > > > i'll get a SEGV. :) > > > > > > MAP_PRIVATE? > > > > > > > This has the undesired effect of me not noticing that the file changed then. > > stat(2) it after the copy? OK, I'm reaching. Handling SEGV seems like a > rather iffy solution at best. In my case, it's a very very very rare thing that the file's been changed out from under me, and I'm doing several thousand copies per second, so doing a stat() on each copy makes things very slow. I know it's a crappy solution, but it's the only one that works. :) Also, consider that *during* the copy, the file could be changed. Kevin To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
Re: SMP + fxp0 wierdness
> Heh... well I've already enabled flags 0xb0ff, which has improved things > somewhat, but our hardware vendor slipped in some IBM DeskStar drives on > us, and they've been no end of trouble. Note that IBM Deskstar drives are very good performers for desktop use (and highly respected), but they are not meant for heavy duty server use. They are *known* to spin down once every 24 hours or so for some kind of head cleaning, and this could certainly be a problem for 24x7 service. Steinar Haug, Nethelp consulting, [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
