Hi all, Currently we have severe problem and are struggling to investigate it. The problem is data broken. Reproduce pocess is following, 0. test FTP function between guest domain and FTP client(other PC). 1. put data from FTP client to guest domain. 2. get data from guest domain to FTP client. 3. compare put data and get data on FTP client. 4. repeat process from step1 to step3. 5. create other guest domain his configuration including vif definition during test repeating. It is more possible to be happen when some FTP tests are running at same time. And at every moments data are broken when FTP client get data from guest domain. When we set SG(scatter/gather) flag off with peth0 of domain0, data are not broken. This problem is happened only particular NIC, 08:04.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5715 Gigabit Ethernet (rev a3) The following NIC case, we can not recreate the problem. 04:00.0 Ethernet controller: Broadcom Corporation NetXtreme II BCM5708S Gigabit Ethernet (rev 11) Do you have any idea for investigating. Thank you. Nishi _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On 20/3/07 07:15, "Hidetoshi Nishi" <nishi.hidetoshi@jp.fujitsu.com> wrote:> It is more possible to be happen when some FTP tests are running at > same time. > And at every moments data are broken when FTP client get data from > guest domain. > When we set SG(scatter/gather) flag off with peth0 of domain0, data > are not broken. > > This problem is happened only particular NIC, > 08:04.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5715 Gigabit > Ethernet (rev a3)Rather worrying. I wonder if you are triggering one of the Broadcom highmem DMA bugs? It would possibly happen that turning off scatter/gather will cause you to copy to skbuffs that happen to be e.g., below 1GB. How much memory does the offending machine have? If more than 1GB, can you repro if you boot Xen with ''mem=1G'' as a boot parameter? How much memory do you allocate to domain0 at boot time? What version of Xen are you using? What version of dom0 kernel? -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Hi Keir, Thank you for reply.> How much memory does the offending machine have?This machine has the memory of 10GB.> If more than 1GB, can you repro if you boot Xen with ''mem=1G'' as a boot > parameter?We tried to boot Xen with ''mem=1G'' as boot parameter. At this time, FTP data was not broken. The environment is as follows. - SG(scatter/gather) flag on - The memory of dom0 is 256MByte. - The memory of guest(PV) domain is 256MByte.> How much memory do you allocate to domain0 at boot time?The memory of dom0 allocated 1GByte at boot time. (dom0_mem=1024M)>What version of Xen are you using?The version of Xen uses RHEL5GA. Therefore, the base of Xen is xen-3.0.3-testing.hg.>What version of dom0 kernel?The version of dom0 is 2.6.18-8. Best regards, -- Takanori Kasai _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
If you are running in PAE mode then you should try modifying the kernel file drivers/net/tg3.c and changing any ''#ifdef CONFIG_HIGHMEM'' to ''#if 0''. It may also be worth doing some tracing around the code that does pci_set_dma_mask() to see what path you are taking there and what mask the driver is choosing to set. -- Keir On 22/3/07 01:16, "Kasai Takanori" <kasai.takanori@jp.fujitsu.com> wrote:> Hi Keir, > > Thank you for reply. > >> How much memory does the offending machine have? > This machine has the memory of 10GB. > >> If more than 1GB, can you repro if you boot Xen with ''mem=1G'' as a boot >> parameter? > We tried to boot Xen with ''mem=1G'' as boot parameter. > At this time, FTP data was not broken. > The environment is as follows. > - SG(scatter/gather) flag on > - The memory of dom0 is 256MByte. > - The memory of guest(PV) domain is 256MByte. > >> How much memory do you allocate to domain0 at boot time? > The memory of dom0 allocated 1GByte at boot time. (dom0_mem=1024M) > >> What version of Xen are you using? > The version of Xen uses RHEL5GA. > Therefore, the base of Xen is xen-3.0.3-testing.hg. > >> What version of dom0 kernel? > The version of dom0 is 2.6.18-8. > > Best regards, > > -- > Takanori Kasai > > > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xensource.com > http://lists.xensource.com/xen-devel_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Hi Keir, Thank you for reply.> If you are running in PAE mode then you should try modifying the kernel file > drivers/net/tg3.c and changing any ''#ifdef CONFIG_HIGHMEM'' to ''#if 0''. It > may also be worth doing some tracing around the code that does > pci_set_dma_mask() to see what path you are taking there and what mask the > driver is choosing to set.However, the same problem happens to another NIC connected with the same machine. It is the following NIC. 05:03.0 Intel Corporation 82545GM Gigabit Ethernet Controller (rev 04) ''# ifdef CONFIG_HIGHMEM'' was not being at least defined by the e1000 driver. When pethX was captured with tcpdump, FTP data was not broken. But, when it was captured in the FTP client(other PC), FTP data was broken. When we create other domain, FTP data seems to be sent again. Do you have any idea for investigating. Best regards, -- Takanori Kasai _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Hi all, We posted following severe problem before. Then we tested this problem under various environments. This problem is happened following NIC, Intel Corporation 82540EM Gigabit Ethernet Controller (rev 02) on Tiger4. It is generated only for the HVM domain of Windows2003 in this environment. Therefore, it seems that it is a potential problem on Xen. This problem is happened following NICs. - Broadcom Corporation NetXtreme BCM5715 Gigabit Ethernet (rev a3) - Intel Corporation 82545GM Gigabit Ethernet Controller (rev 04) - Intel Corporation 82540EM Gigabit Ethernet Controller (rev 02) The following NIC cases, we can not create the problem. - Broadcom Corporation NetXtreme II BCM5708S Gigabit Ethernet (rev 11) - Broadcom Corporation NetXtreme BCM5701 Gigabit Ethernet (rev 15) - Intel Corporation 82546GB Gigabit Ethernet Controller (rev 03) We think that this problem is serious. When pethX was captured, FTP data was not broken. But, when it was captured in the FTP client(other PC), FTP data was broken. When we create other domain, FTP data seems to be sent again. It seems that it happens by NIC in which the offload function is provided. Could you test other environments? We offer you the FTP test. (ftp_test.tar.gz) Best regards, -- Takanori Kasai ----- Original Message ----- From: "Hidetoshi Nishi" <nishi.hidetoshi@jp.fujitsu.com> To: <xen-devel@lists.xensource.com> Sent: Tuesday, March 20, 2007 4:15 PM Subject: [Xen-devel] Data broken during FTP test> > Hi all, > > Currently we have severe problem and are struggling to investigate it. > The problem is data broken. > > Reproduce pocess is following, > 0. test FTP function between guest domain and FTP client(other PC). > 1. put data from FTP client to guest domain. > 2. get data from guest domain to FTP client. > 3. compare put data and get data on FTP client. > 4. repeat process from step1 to step3. > 5. create other guest domain his configuration including vif > definition during test repeating. > > It is more possible to be happen when some FTP tests are running at > same time. > And at every moments data are broken when FTP client get data from > guest domain. > When we set SG(scatter/gather) flag off with peth0 of domain0, data > are not broken. > > This problem is happened only particular NIC, > 08:04.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5715 Gigabit > Ethernet (rev a3) > > The following NIC case, we can not recreate the problem. > 04:00.0 Ethernet controller: Broadcom Corporation NetXtreme II BCM5708S > Gigabit > Ethernet (rev 11) > > Do you have any idea for investigating._______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
> When pethX was captured, FTP data was not broken. > But, when it was captured in the FTP client(other PC), FTP data was > broken. > When we create other domain, FTP data seems to be sent again. > It seems that it happens by NIC in which the offload function is > provided.Please can you provide more information about the exact xen version used, and the test configuration. What is the nature of the corruption introduced? Does using ethtool to turn csum offload off on either the peth or vif make any difference? Ian _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Hi Ian, Thank you for reply.> Please can you provide more information about the exact xen version > used, and the test configuration. What is the nature of the corruption > introduced?The version of Xen used RHEL5GA. This problem has happened comparatively only by new NIC. It is following NIC that this problem happened first. Broadcom Corporation NetXtreme BCM5715 Gigabit Ethernet (rev a3) The PV domain(domain_1) is created, and FTP test jobs were executed between PV domain and FTP client(other PC). Other PV domain(domain_2) connected with the same bridge are created. The FTP data was broken when the network started on other created PV domains. FTP data was broken only when it is transmitted from PV domain to FTP client. We offer the material of environment and configuration files. However, this problem doesn''t happen because of the change in the setting. 1. SG(scatter/gather) flag off 2. ''mem=1G'' is set as a boot parameter, and SG(scatter/gather) flag on. We are guessing this problem as follows.(Data_broken_during_FTP_test.doc) Frame information and data are collected by Scatter/Gather offload function on the NIC side and frame is made. The buffer data of Dom0 is liberated with ack from NIC. Therefore, the buffer is used other forwarding. The request of sending again is notified to guest OS when there is a request of the TCP sending again. However, it seems to read data from the buffer because data is maintained in the buffer of Dom0 as for NIC and to send it again. When the SG flag is turning off, the request of the TCP sending again is notified and event guest OS is notified. Therefore, FTP data is not broken because it is sent again by guest OS(domain_1). However, the reason why this problem doesn''t happen when the memory of Xen is limited is not understood. Best Regards, -- Takanori Kasai _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On 18/4/07 11:57, "Kasai Takanori" <kasai.takanori@jp.fujitsu.com> wrote:> However, this problem doesn''t happen because of the change in the setting. > 1. SG(scatter/gather) flag off > 2. ''mem=1G'' is set as a boot parameter, and SG(scatter/gather) flag on.This might indicate a high-memory bug in the net driver(s). The TCP re-send theory makes no sense -- the decision to re-send is always made by the TCP protocol stack in the domU. dom0 itself will never decide to re-send. How much memory does your test system have? Are you running 64-bit Xen, PAE Xen, or non-PAE 32-bit Xen? -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On 18/4/07 03:06, "Kasai Takanori" <kasai.takanori@jp.fujitsu.com> wrote:> This problem is happened following NIC, > Intel Corporation 82540EM Gigabit Ethernet Controller (rev 02) on Tiger4. > It is generated only for the HVM domain of Windows2003 in this environment. > Therefore, it seems that it is a potential problem on Xen.Have you seen this problem only with HVM domU guests? If so, presumably you are using emulated network I/O? -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Hi Keir, Thank you for reply.>> However, this problem doesn''t happen because of the change in the setting. >> 1. SG(scatter/gather) flag off >> 2. ''mem=1G'' is set as a boot parameter, and SG(scatter/gather) flag on. > > This might indicate a high-memory bug in the net driver(s). The TCP re-send > theory makes no sense -- the decision to re-send is always made by the TCP > protocol stack in the domU. dom0 itself will never decide to re-send. > > How much memory does your test system have? Are you running 64-bit Xen, PAE > Xen, or non-PAE 32-bit Xen?The environment of the machine is as follows. Machine : PRIMERGY model RX200S3 CPU : Xeon 5160 3.0 GHz * 2 Memory : 12 GB(2048 MB * 6) HDD : AL-9LX 73GB * 2 (Onboard IME RAID1) NIC : Broadcom Corporation NetXtreme BC5715 Gigabit Ethernet (rev a3) * 2 Intel Corporation 82545GM Gigabit Ethernet Controller (rev 04) * 1 OS : Red Hat Enterprise Linux 5 GA It is confirmed that this problem happens similarly on NIC of the Intel Corporation. Best Regards, -- Takanori Kasai _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
> >> However, this problem doesn''t happen because of the change in the > setting. > >> 1. SG(scatter/gather) flag off > >> 2. ''mem=1G'' is set as a boot parameter, and SG(scatter/gather) flag > on.Is this is a 32b install of RHEL5 GA, or x86_64? I presume 32b. Can you reproduce if you uninstall the RHEL5 xen/kernel and xen tools and libs RPMs and then install binaries built from xen-unstable? Rather than disabling SG, what happens if you disable TSO? What about checksum offload (TX and RX)? What happens with mem=4G? This is a worrying bug -- thanks for your help in tracking it down. Thanks, Ian> > This might indicate a high-memory bug in the net driver(s). The TCP > re-send > > theory makes no sense -- the decision to re-send is always made bythe> TCP > > protocol stack in the domU. dom0 itself will never decide tore-send.> > > > How much memory does your test system have? Are you running 64-bit > Xen, PAE > > Xen, or non-PAE 32-bit Xen? > > The environment of the machine is as follows. > > Machine : PRIMERGY model RX200S3 > CPU : Xeon 5160 3.0 GHz * 2 > Memory : 12 GB(2048 MB * 6) > HDD : AL-9LX 73GB * 2 (Onboard IME RAID1) > NIC : Broadcom Corporation NetXtreme BC5715 Gigabit Ethernet (rev > a3) * 2 > Intel Corporation 82545GM Gigabit Ethernet Controller (rev04)> * 1 > OS : Red Hat Enterprise Linux 5 GA > > It is confirmed that this problem happens similarly on NIC of theIntel> Corporation. > > Best Regards, > > -- > Takanori Kasai >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Hi Ian, Thank you for reply.> Is this is a 32b install of RHEL5 GA, or x86_64? I presume 32b.As for this environment, 32bit version of RHEL5GA is installed.> Can you reproduce if you uninstall the RHEL5 xen/kernel and xen tools > and libs RPMs and then install binaries built from xen-unstable?We will reproduce it with xen-unstable.hg.> Rather than disabling SG, what happens if you disable TSO?When TSO is disabling, FTP data was broken. (enabling SG) When SG is disabling, TSO automatically is disabling, too. TSO SG Reproducibility ------------+-----------+---------------------- Enable Enable Broken Enable Disable (Impossibility) Disable Enable Broken Disable Disable Not broken> What about checksum offload (TX and RX)?TX and RX are enabling. When TX and RX are disabling, reproducibility didn''t change.> What happens with mem=4G?We will reproduce it. Best Regards, -- Takanori Kasai _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Hi Ian and Keir, I''m sorry for delaying response.>> Can you reproduce if you uninstall the RHEL5 xen/kernel and xen tools >> and libs RPMs and then install binaries built from xen-unstable? > > We will reproduce it with xen-unstable.hg.We tested it with the same machine. The environment is as follows. Xen : - xen-unstable.hg: cs15072 NICs : - Broadcom Corporation NetXtreme BCM5715 Gigabit Ethernet (rev a3) - Intel Corporation 82545GM Gigabit Ethenet Controller (rev 04) Guests : - DomU - DomVTi(RHEL5GA) - DomVTi(RHEL5GA) + PV_ON_HVM(VNIF) Others: - SG and TSO flag: ENABLE - Memory ・Domain0 : 1GByte ・Guest Domains : 2GByte The test result was not reproduced as follows. | Broadcom | Intel ----------------------------+---------------+--------------- DomU | Not broken | Not broken DomVTi(RHEL5GA) | Not broken | Not broken DomVTi(RHEL5GA)+VNIF | Not broken | Not broken Next, the tg3 driver and the e1000 driver were replaced with RHEL5GA. However, the result was not reproduced similarly. Is there a correction related to this problem since xen3.0.3? Best Regards, -- Takanori Kasai ----- Original Message ----- From: "Kasai Takanori" <kasai.takanori@jp.fujitsu.com> To: "Ian Pratt" <Ian.Pratt@cl.cam.ac.uk>; "Keir Fraser" <Keir.Fraser@cl.cam.ac.uk>; "xen-devel" <xen-devel@lists.xensource.com> Sent: Friday, April 20, 2007 11:03 AM Subject: Re: [Xen-devel] Data broken during FTP test> Hi Ian, > > Thank you for reply. > >> Is this is a 32b install of RHEL5 GA, or x86_64? I presume 32b. > As for this environment, 32bit version of RHEL5GA is installed. >> Can you reproduce if you uninstall the RHEL5 xen/kernel and xen tools >> and libs RPMs and then install binaries built from xen-unstable? > We will reproduce it with xen-unstable.hg. >> Rather than disabling SG, what happens if you disable TSO? > When TSO is disabling, FTP data was broken. (enabling SG) > When SG is disabling, TSO automatically is disabling, too. > TSO SG Reproducibility > ------------+-----------+---------------------- > Enable Enable Broken > Enable Disable (Impossibility) > Disable Enable Broken > Disable Disable Not broken > >> What about checksum offload (TX and RX)? > TX and RX are enabling. When TX and RX are disabling, reproducibility didn''t > change. >> What happens with mem=4G? > We will reproduce it. > Best Regards, > > -- > Takanori Kasai > > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xensource.com > http://lists.xensource.com/xen-devel >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Hi All, I''m sorry the mistake was found in mail. It is a mistake of not DomVTi but DomVTx. Best Regards, -- Takanori Kasai ----- Original Message ----- From: "Kasai Takanori" <kasai.takanori@jp.fujitsu.com> To: "Ian Pratt" <Ian.Pratt@cl.cam.ac.uk>; "Keir Fraser" <Keir.Fraser@cl.cam.ac.uk> Cc: "xen-devel" <xen-devel@lists.xensource.com> Sent: Thursday, May 24, 2007 6:33 PM Subject: Re: [Xen-devel] Data broken during FTP test> Hi Ian and Keir, > > I''m sorry for delaying response. > > >>> Can you reproduce if you uninstall the RHEL5 xen/kernel and xen tools >>> and libs RPMs and then install binaries built from xen-unstable? >> >> We will reproduce it with xen-unstable.hg. > > We tested it with the same machine. > The environment is as follows. > > Xen : > - xen-unstable.hg: cs15072 > NICs : > - Broadcom Corporation NetXtreme BCM5715 Gigabit Ethernet (rev a3) > - Intel Corporation 82545GM Gigabit Ethenet Controller (rev 04) > Guests : > - DomU > - DomVTi(RHEL5GA) > - DomVTi(RHEL5GA) + PV_ON_HVM(VNIF) > Others: > - SG and TSO flag: ENABLE > - Memory > ・Domain0 : 1GByte > ・Guest Domains : 2GByte > > The test result was not reproduced as follows. > > | Broadcom | Intel > ----------------------------+---------------+--------------- > DomU | Not broken | Not broken > DomVTi(RHEL5GA) | Not broken | Not broken > DomVTi(RHEL5GA)+VNIF | Not broken | Not broken > > Next, the tg3 driver and the e1000 driver were replaced with RHEL5GA. > However, the result was not reproduced similarly. > > Is there a correction related to this problem since xen3.0.3? > > Best Regards, > > -- > Takanori Kasai_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Hi All, We are investigating the problem that data was broken during FTP test. It did not reproduce it in the latest changeset. (xen-unstable.hg: cs15072) We tested other changesets. The test result is as follows. version | result -----------------------+--------------- xen-3.0.3-testing.hg | Broken xen-3.0.4-testing.hg | Not Broken xen-3.1-testing.hg | Not Broken Therefore, it seems that it is corrected between from 3.0.3 to 3.0.4. We will investigate by which changeset this problem is corrected. Best Regards, -- Takanori Kasai ----- Original Message ----- From: "Kasai Takanori" <kasai.takanori@jp.fujitsu.com> To: "Ian Pratt" <Ian.Pratt@cl.cam.ac.uk>; "Keir Fraser" <Keir.Fraser@cl.cam.ac.uk> Cc: "xen-devel" <xen-devel@lists.xensource.com> Sent: Thursday, May 24, 2007 8:04 PM Subject: Re: [Xen-devel] Data broken during FTP test> Hi All, > > I''m sorry the mistake was found in mail. > It is a mistake of not DomVTi but DomVTx. > > Best Regards, > > -- > Takanori Kasai > > ----- Original Message ----- > From: "Kasai Takanori" <kasai.takanori@jp.fujitsu.com> > To: "Ian Pratt" <Ian.Pratt@cl.cam.ac.uk>; "Keir Fraser" > <Keir.Fraser@cl.cam.ac.uk> > Cc: "xen-devel" <xen-devel@lists.xensource.com> > Sent: Thursday, May 24, 2007 6:33 PM > Subject: Re: [Xen-devel] Data broken during FTP test > > >> Hi Ian and Keir, >> >> I''m sorry for delaying response. >> >> >>>> Can you reproduce if you uninstall the RHEL5 xen/kernel and xen tools >>>> and libs RPMs and then install binaries built from xen-unstable? >>> >>> We will reproduce it with xen-unstable.hg. >> >> We tested it with the same machine. >> The environment is as follows. >> >> Xen : >> - xen-unstable.hg: cs15072 >> NICs : >> - Broadcom Corporation NetXtreme BCM5715 Gigabit Ethernet (rev a3) >> - Intel Corporation 82545GM Gigabit Ethenet Controller (rev 04) >> Guests : >> - DomU >> - DomVTi(RHEL5GA) >> - DomVTi(RHEL5GA) + PV_ON_HVM(VNIF) >> Others: >> - SG and TSO flag: ENABLE >> - Memory >> ・Domain0 : 1GByte >> ・Guest Domains : 2GByte >> >> The test result was not reproduced as follows. >> >> | Broadcom | Intel >> ----------------------------+---------------+--------------- >> DomU | Not broken | Not broken >> DomVTi(RHEL5GA) | Not broken | Not broken >> DomVTi(RHEL5GA)+VNIF | Not broken | Not broken >> >> Next, the tg3 driver and the e1000 driver were replaced with RHEL5GA. >> However, the result was not reproduced similarly. >> >> Is there a correction related to this problem since xen3.0.3? >> >> Best Regards, >> >> -- >> Takanori Kasai > > > > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xensource.com > http://lists.xensource.com/xen-devel >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Hi All, We are investigating the problem that data was broken during FTP test. This problem seems not to be reproduced since changeset 13511 of xen-unstable.hg. Moreover, it is generated only by PAE of X86. (It did not reproduce it when working with x86_64 in the same machine.) When HIGHMEM was not used, it didn''t reproduce it. The kernel has been updated from 2.6.17 to 2.6.18 by this changeset. Because the amount of the correction is large, we do not understand where of this change set this problem is corrected. Will you understand where to be corrected? If the part where this problem is corrected can be understood, could you teach to us? The environment for your information reproduced is as follows. Xen : - xen-unstable.hg: before cs13511 NICs : - Broadcom Corporation NetXtreme BCM5715 Gigabit Ethernet (rev a3) - Intel Corporation 82545GM Gigabit Ethenet Controller (rev 04) Guests : - DomU - DomVTx(RHEL5GA) - DomVTx(RHEL5GA) + PV_ON_HVM(VNIF) Others: - SG and TSO flag: ENABLE - Memory ・Domain0 : HIGHMEM ・Guest Domains : 2GByte Best Regards, -- Takanori Kasai ----- Original Message ----- From: "Kasai Takanori" <kasai.takanori@jp.fujitsu.com> To: "Ian Pratt" <Ian.Pratt@cl.cam.ac.uk>; "Keir Fraser" <Keir.Fraser@cl.cam.ac.uk> Cc: "xen-devel" <xen-devel@lists.xensource.com> Sent: Tuesday, June 05, 2007 8:00 PM Subject: Re: [Xen-devel] Data broken during FTP test> Hi All, > > We are investigating the problem that data was broken during FTP test. > > It did not reproduce it in the latest changeset. (xen-unstable.hg: cs15072) > > We tested other changesets. > The test result is as follows. > > version | result > -----------------------+--------------- > xen-3.0.3-testing.hg | Broken > xen-3.0.4-testing.hg | Not Broken > xen-3.1-testing.hg | Not Broken > > Therefore, it seems that it is corrected between from 3.0.3 to 3.0.4. > We will investigate by which changeset this problem is corrected. > > Best Regards, > > -- > Takanori Kasai > > > ----- Original Message ----- > From: "Kasai Takanori" <kasai.takanori@jp.fujitsu.com> > To: "Ian Pratt" <Ian.Pratt@cl.cam.ac.uk>; "Keir Fraser" > <Keir.Fraser@cl.cam.ac.uk> > Cc: "xen-devel" <xen-devel@lists.xensource.com> > Sent: Thursday, May 24, 2007 8:04 PM > Subject: Re: [Xen-devel] Data broken during FTP test > > >> Hi All, >> >> I''m sorry the mistake was found in mail. >> It is a mistake of not DomVTi but DomVTx. >> >> Best Regards, >> >> -- >> Takanori Kasai >> >> ----- Original Message ----- >> From: "Kasai Takanori" <kasai.takanori@jp.fujitsu.com> >> To: "Ian Pratt" <Ian.Pratt@cl.cam.ac.uk>; "Keir Fraser" >> <Keir.Fraser@cl.cam.ac.uk> >> Cc: "xen-devel" <xen-devel@lists.xensource.com> >> Sent: Thursday, May 24, 2007 6:33 PM >> Subject: Re: [Xen-devel] Data broken during FTP test >> >> >>> Hi Ian and Keir, >>> >>> I''m sorry for delaying response. >>> >>> >>>>> Can you reproduce if you uninstall the RHEL5 xen/kernel and xen tools >>>>> and libs RPMs and then install binaries built from xen-unstable? >>>> >>>> We will reproduce it with xen-unstable.hg. >>> >>> We tested it with the same machine. >>> The environment is as follows. >>> >>> Xen : >>> - xen-unstable.hg: cs15072 >>> NICs : >>> - Broadcom Corporation NetXtreme BCM5715 Gigabit Ethernet (rev a3) >>> - Intel Corporation 82545GM Gigabit Ethenet Controller (rev 04) >>> Guests : >>> - DomU >>> - DomVTi(RHEL5GA) >>> - DomVTi(RHEL5GA) + PV_ON_HVM(VNIF) >>> Others: >>> - SG and TSO flag: ENABLE >>> - Memory >>> ・Domain0 : 1GByte >>> ・Guest Domains : 2GByte >>> >>> The test result was not reproduced as follows. >>> >>> | Broadcom | Intel >>> ----------------------------+---------------+--------------- >>> DomU | Not broken | Not broken >>> DomVTi(RHEL5GA) | Not broken | Not broken >>> DomVTi(RHEL5GA)+VNIF | Not broken | Not broken >>> >>> Next, the tg3 driver and the e1000 driver were replaced with RHEL5GA. >>> However, the result was not reproduced similarly. >>> >>> Is there a correction related to this problem since xen3.0.3? >>> >>> Best Regards, >>> >>> -- >>> Takanori Kasai >> >> >> >> >> _______________________________________________ >> Xen-devel mailing list >> Xen-devel@lists.xensource.com >> http://lists.xensource.com/xen-devel >> > > > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xensource.com > http://lists.xensource.com/xen-devel >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Seemingly Similar Threads
- [PATCH] [IOEMU] Allow blktap to be able to be booted as systemvolume for PV-on-HVM(TAKE 3)
- [PATCH] Fix keymap for Japanese keyboard
- [PATCH]Add the Japanese keymap for VNC Server.
- [PATCH] qemu-dm monitor doesn''t support the quit command
- [PATCH] Fix change of CDROM for block-configure command