From royfran at paran.com Sun Jul 1 22:39:39 2007 From: royfran at paran.com (Young ju Kim) Date: Sun, 01 Jul 2007 22:39:39 PDT Subject: [ethbridge-discuss] when is the patch plan? Message-ID: <30264284.1183354809786.JavaMail.Twebapp@oss-app1> Hi! It has an interest in this project. Makes IPS of solaris, bases it wants using this technique of the hazard which. But currently version is problem. Version of present time there is a fatal bug which gets up panic. With ipfilter gearing when will become the bedspread? It wants knowing the plan against a new patch version. If is not plan and it entrusts the advice which is different method. It will wait a reply. It thanks. This message posted from opensolaris.org From ford at omnicron.com Mon Jul 9 11:09:20 2007 From: ford at omnicron.com (Mike "Ford" Ditto) Date: Mon, 09 Jul 2007 11:09:20 -0700 Subject: [ethbridge-discuss] when is the patch plan? In-Reply-To: <30264284.1183354809786.JavaMail.Twebapp@oss-app1> (message from Young ju Kim on Sun, 01 Jul 2007 22:39:39 PDT) References: <30264284.1183354809786.JavaMail.Twebapp@oss-app1> Message-ID: <469279D0.2F45@han.omnicron.com> Young ju Kim, If you post some info about the panic you found maybe someone can tell you if it is a known problem or if there is a fix. I know of a panic on SPARC systems when bridge_debug is on, and another panic when deleting an interface from a bridge or deleting a bridge. Dong-Hai Han was working on a more advanced version of the code, but he was concentrating on finishing the spanning tree protocol and management functionality, so I don't know if he spent much effort on the basic reliability issues - it was still "prototype quality" last I heard. But we can encourage him to post what he has. -=] Mike [=- From royfran at paran.com Mon Jul 9 19:18:17 2007 From: royfran at paran.com (Young ju Kim) Date: Mon, 09 Jul 2007 19:18:17 PDT Subject: [ethbridge-discuss] when is the patch plan? In-Reply-To: <30264284.1183354809786.JavaMail.Twebapp@oss-app1> Message-ID: <29540220.1184033956659.JavaMail.Twebapp@oss-app1> Hi. Mike. Thanks for your sincere reply. We are currently using ?Solaris x86 Update 3? which support Intel based 10/100/1000 LAN card(e1000g). Unfortunately this system produces ?Panic Condition? on our system. It also does not transfer data that lager than MTU Size. Moreover, it does not connect to ITF filter. It wants solving inside the time when it is quick our problem. Also against the version which it is developing now it wants test. Could you pleas give us Mr. Han?s e-mail address that we can contact? Best regards, Roy Kim This message posted from opensolaris.org From Donghai.Han at Sun.COM Wed Jul 11 02:33:51 2007 From: Donghai.Han at Sun.COM (Dong-Hai Han) Date: Wed, 11 Jul 2007 17:33:51 +0800 Subject: [ethbridge-discuss] when is the patch plan? Message-ID: <4694A3FF.8000006@Sun.COM> > We are currently using ?Solaris x86 Update 3? which support Intel based 10/100/1000 LAN card(e1000g). > Unfortunately this system produces ?Panic Condition? on our system. It also does not transfer data that lager than MTU Size. Moreover, it does not connect to ITF filter. > It wants solving inside the time when it is quick our problem. Also against the version which it is developing now it wants test. > Could you pleas give us Mr. Han?s e-mail address that we can contact? Could you please give us more information about your panic, like how it is triggered, and the call stack? Best, Donghai. From royfran at paran.com Thu Jul 12 18:48:24 2007 From: royfran at paran.com (Young ju Kim) Date: Thu, 12 Jul 2007 18:48:24 PDT Subject: [ethbridge-discuss] when is the patch plan? In-Reply-To: <4694A3FF.8000006@Sun.COM> Message-ID: <8301437.1184291334533.JavaMail.Twebapp@oss-app1> Hi. Dong Hai Han. The panic the actual condition with the writings which come up to former times is same. That it sees that it is more urgent, the data above MTU size and it means that does not become the communication. Also it is a problem where the ping back is not accomplished with the server which is connected with an equipment from the equipment which becomes the bridge. The iprb from the driver the communication of big size is become accomplished communication of big size does not become accomplished from e1000g driver. Please, it gives a solution plan and it wishes. Best regards, Young Ju Kim This message posted from opensolaris.org From Donghai.Han at Sun.COM Fri Jul 13 03:50:11 2007 From: Donghai.Han at Sun.COM (Dong-Hai Han) Date: Fri, 13 Jul 2007 18:50:11 +0800 Subject: [ethbridge-discuss] when is the patch plan? In-Reply-To: <4694A3FF.8000006@Sun.COM> References: <4694A3FF.8000006@Sun.COM> Message-ID: <469758E3.2070109@Sun.COM> Hello, Young Ju, (I am not on the alias, so please use reply-all) Do you mean that you have a network setup that works with iprb driver, but will fail with e1000g driver? It would be helpful if you tell us more detail about your setup, and how you generated "the data above MTU size". Best, Donghai. Young Ju Kim Wrote: > The panic the actual condition with the writings which come up to former times is same. > That it sees that it is more urgent, the data above MTU size and it means that does not become the communication. Also it is a problem where the ping back is not accomplished with the server which is connected with an equipment from the equipment which becomes the bridge. > The iprb from the driver the communication of big size is become accomplished communication of big size does not become accomplished from e1000g driver. > Please, it gives a solution plan and it wishes. From ruedigerteichert at msi.com.tw Tue Jul 17 02:00:14 2007 From: ruedigerteichert at msi.com.tw (Ruediger Teichert) Date: Tue, 17 Jul 2007 02:00:14 PDT Subject: [ethbridge-discuss] HTTP not going through Message-ID: <31581352.1184662872079.JavaMail.Twebapp@oss-app1> Dears, hope I can post a question here. I have an AMD64 system with Solaris 10 and have setup the bridge. My Solaris box has a bridge consisting of 3 NICs (Gigalan), one called e1000g5 is configured as DHCP client (ADSL router "output" line connected, the Solaris box can go to Internet with it), the other two called e1000g6 and e1000g7 are configured with NULL IP (0.0.0.0) and are up. I then connected e1000g6 and e1000g7 to Windows Notebook and PC (both being configured to dynamic IPs) and... (1). It basically works. Both Windows boxes are getting their IPs from my ADSL router. Great. (2). I open a browser (not matter Firefox or IE) and type an URL and... first everything works fine, "website found, waiting for answer", but it never loads. Endless waiting for website until finally IE displays the usual error message, website cannot be displayed. What am I doing wrong? Any help greatly appreciated, thanks! Ruediger This message posted from opensolaris.org From dme at sun.com Tue Jul 17 02:48:57 2007 From: dme at sun.com (David Edmondson) Date: Tue, 17 Jul 2007 10:48:57 +0100 Subject: [ethbridge-discuss] HTTP not going through In-Reply-To: <31581352.1184662872079.JavaMail.Twebapp@oss-app1> References: <31581352.1184662872079.JavaMail.Twebapp@oss-app1> Message-ID: <20070717094857.GB31250@apfelstrudel.hh.sledj.net> On Tue, Jul 17, 2007 at 02:00:14AM -0700, Ruediger Teichert wrote: > I have an AMD64 system with Solaris 10 and have setup the bridge. My > Solaris box has a bridge consisting of 3 NICs (Gigalan), one called > e1000g5 is configured as DHCP client (ADSL router "output" line > connected, the Solaris box can go to Internet with it), the other > two called e1000g6 and e1000g7 are configured with NULL IP (0.0.0.0) > and are up. I then connected e1000g6 and e1000g7 to Windows > Notebook and PC (both being configured to dynamic IPs) and... > > (1). It basically works. Both Windows boxes are getting their IPs > from my ADSL router. Great. > (2). I open a browser (not matter Firefox or IE) and type an URL > and... first everything works fine, "website found, waiting for > answer", but it never loads. Endless waiting for website until > finally IE displays the usual error message, website cannot be > displayed. > > What am I doing wrong? Any help greatly appreciated, thanks! There is a bug in the e1000g driver (6490623) where received packets are passed upstream with the ethernet CRC attached. This can result in packets that are longer than a legal ethernet frame, a fact that the bridge code finds upsetting. As a result it drops the packets. So, if you are receiving full frames (common during bulk data transfer, like a web page download) those frames get dropped. You can test if this is what is happening by running the following dtrace script on the bridge machine: #!/usr/sbin/dtrace -s fbt:mac:mac_rx:entry /(((mblk_t *)arg2)->b_wptr - ((mblk_t *)arg2)->b_rptr) > 1514/ { trace((((mblk_t *)arg2)->b_wptr - ((mblk_t *)arg2)->b_rptr)); stack(1); } If you see output then the bug is biting. It is fixed in Nevada build 69. dme. From ruedigerteichert at msi.com.tw Tue Jul 17 03:50:19 2007 From: ruedigerteichert at msi.com.tw (Ruediger Teichert) Date: Tue, 17 Jul 2007 03:50:19 PDT Subject: [ethbridge-discuss] HTTP not going through In-Reply-To: <20070717094857.GB31250@apfelstrudel.hh.sledj.net> Message-ID: <31259809.1184669449207.JavaMail.Twebapp@oss-app1> Thanks for the info. I have started this as a shell script and get the output "dtrace: script 'test' matched 1 probe. With 'test' being the script. So I guess I have to update Solaris. This message posted from opensolaris.org From dme at sun.com Tue Jul 17 03:53:42 2007 From: dme at sun.com (David Edmondson) Date: Tue, 17 Jul 2007 11:53:42 +0100 Subject: [ethbridge-discuss] HTTP not going through In-Reply-To: <31259809.1184669449207.JavaMail.Twebapp@oss-app1> References: <20070717094857.GB31250@apfelstrudel.hh.sledj.net> <31259809.1184669449207.JavaMail.Twebapp@oss-app1> Message-ID: <20070717105341.GD4692@sacher-torte.uk.sun.com> On Tue, Jul 17, 2007 at 03:50:19AM -0700, Ruediger Teichert wrote: > Thanks for the info. I have started this as a shell script and get the output "dtrace: script 'test' matched 1 probe. With 'test' being the script. So I guess I have to update Solaris. Sorry, no, that's not the output I meant :-) The "matched 1 probe" means that the script is running. Leave it running while you try to browse the web from your Windows machine. It should output more data. dme. From ruedigerteichert at msi.com.tw Tue Jul 17 18:07:35 2007 From: ruedigerteichert at msi.com.tw (Ruediger Teichert) Date: Tue, 17 Jul 2007 18:07:35 PDT Subject: [ethbridge-discuss] HTTP not going through In-Reply-To: <20070717105341.GD4692@sacher-torte.uk.sun.com> Message-ID: <18507612.1184720885673.JavaMail.Twebapp@oss-app1> Ups, sorry :-) I got more output soon: CPU ID FUNCTION:NAME 3 25673 mac_rx:entry 1518 e1000g' e1000g_intr_work+0x146 repeating all over again. This message posted from opensolaris.org From dme at sun.com Tue Jul 17 22:56:42 2007 From: dme at sun.com (David Edmondson) Date: Wed, 18 Jul 2007 06:56:42 +0100 Subject: [ethbridge-discuss] HTTP not going through In-Reply-To: <18507612.1184720885673.JavaMail.Twebapp@oss-app1> References: <20070717105341.GD4692@sacher-torte.uk.sun.com> <18507612.1184720885673.JavaMail.Twebapp@oss-app1> Message-ID: <20070718055642.GB8064@sacher-torte.uk.sun.com> On Tue, Jul 17, 2007 at 06:07:35PM -0700, Ruediger Teichert wrote: > Ups, sorry :-) I got more output soon: > > CPU ID FUNCTION:NAME > 3 25673 mac_rx:entry 1518 e1000g' e1000g_intr_work+0x146 > > repeating all over again. This looks like the bug, though I mis-remembered the details (it's not the bridge module that drops the packet, rather it's the transmit path once the bridge has forwarded the packet on). If you want this to work then one option is to check for packets longer than 1514 bytes in the bridge forwarding code and truncate them (to 1514 bytes). This workaround was useful for me in the past. The underlying (e1000g) bug is now fixed in Nevada, likely scheduled for inclusion in S10U5. dme. From ruedigerteichert at msi.com.tw Mon Jul 23 02:12:56 2007 From: ruedigerteichert at msi.com.tw (Ruediger Teichert) Date: Mon, 23 Jul 2007 02:12:56 PDT Subject: [ethbridge-discuss] HTTP not going through In-Reply-To: <20070718055642.GB8064@sacher-torte.uk.sun.com> Message-ID: <26987396.1185182006339.JavaMail.Twebapp@oss-app1> Thanks. Is there a suitable patch or likewise? Or do you happen to know when Nevada Build 69 is available? Otherwise a hint where to change the code for the truncation would be nice ... This message posted from opensolaris.org From dme at sun.com Mon Jul 23 03:21:58 2007 From: dme at sun.com (David Edmondson) Date: Mon, 23 Jul 2007 11:21:58 +0100 Subject: [ethbridge-discuss] HTTP not going through In-Reply-To: <26987396.1185182006339.JavaMail.Twebapp@oss-app1> References: <26987396.1185182006339.JavaMail.Twebapp@oss-app1> Message-ID: <1185186118.4759.3.camel@sacher-torte.hh.sledj.net> On Mon, 2007-07-23 at 02:12 -0700, Ruediger Teichert wrote: > Thanks. Is there a suitable patch or likewise? Or do you happen to > know when Nevada Build 69 is available? I'm not sure when SXCE for build 69 will be available, sorry. > Otherwise a hint where to change the code for the truncation would be > nice ... The attached patch is untested, but is basically the same as something that we used a while ago. dme. -------------- next part -------------- A non-text attachment was scrubbed... Name: streams.c.diff Type: text/x-patch Size: 510 bytes Desc: not available URL: From wm.jiang at 126.com Fri Jul 27 00:36:46 2007 From: wm.jiang at 126.com (Eric jiang) Date: Fri, 27 Jul 2007 00:36:46 PDT Subject: [ethbridge-discuss] brctl: open(/devices/pseudo/bridge@0:0) failed Message-ID: <1969201.1185521836626.JavaMail.Twebapp@oss-app1> when i create mybdg as #brctl addbr mybdg output is brctl: open(/devices/pseudo/bridge at 0:0) failed: No such file or directory what's mean? thanks for you! This message posted from opensolaris.org From ruedigerteichert at msi.com.tw Fri Jul 27 00:48:17 2007 From: ruedigerteichert at msi.com.tw (Ruediger Teichert) Date: Fri, 27 Jul 2007 00:48:17 PDT Subject: [ethbridge-discuss] brctl: open(/devices/pseudo/bridge@0:0) failed In-Reply-To: <1969201.1185521836626.JavaMail.Twebapp@oss-app1> Message-ID: <8827542.1185522527209.JavaMail.Twebapp@oss-app1> I had that too once. I think I had forgotten the # add_drv bridge step, which is not mentioned in the README or HOWTO. You may want to check if you did this step. This message posted from opensolaris.org From wm.jiang at 126.com Sun Jul 29 22:06:32 2007 From: wm.jiang at 126.com (Eric jiang) Date: Sun, 29 Jul 2007 22:06:32 PDT Subject: [ethbridge-discuss] brctl: open(/devices/pseudo/bridge@0:0) failed In-Reply-To: <8827542.1185522527209.JavaMail.Twebapp@oss-app1> Message-ID: <149726.1185772022400.JavaMail.Twebapp@oss-app1> Thank you very march! It's running! This message posted from opensolaris.org From wm.jiang at 126.com Sun Jul 29 22:39:35 2007 From: wm.jiang at 126.com (Eric jiang) Date: Sun, 29 Jul 2007 22:39:35 PDT Subject: [ethbridge-discuss] Can't delete bridge Message-ID: <5295142.1185774005812.JavaMail.Twebapp@oss-app1> I can't delete mybdg bridge. bash-3.00#brctl show mybdg Bridge mybdg is /devices/pseudo/bridge at 0:0 bridge name bridge id STP enabled interfaces mybdg ? no pcn0 bash-3.00#brctl delif mybdg pcn0 Bridge mybdg is /deices/pseudo/bridge at 0:0 brctl:Removed pcn0 from mybdg bash-3.00#brctl delbr mydbg brctl:Can't delete bridge mybdg: Device busy How do it? or it's a bug? Thanks This message posted from opensolaris.org From ruedigerteichert at msi.com.tw Sun Jul 29 22:57:09 2007 From: ruedigerteichert at msi.com.tw (Ruediger Teichert) Date: Sun, 29 Jul 2007 22:57:09 PDT Subject: [ethbridge-discuss] Can't delete bridge In-Reply-To: <5295142.1185774005812.JavaMail.Twebapp@oss-app1> Message-ID: <9322566.1185775059652.JavaMail.Twebapp@oss-app1> Well, the Errata thread here mentions: (3) There is a bug in the code that can cause a panic when deleting an interface from a bridge or when deleting a bridge. Nobody has investigated this in detail yet. Don't delete interfaces unless you can handle a panic. But you have to delete the interfaces before you can delete the bridge. But deleting interfaces doesn't really work (see above). I tried but couldn't delete the bridge. After restart it is gone though :-() This message posted from opensolaris.org