> > From: Matt Fleming [matt at console-pimps.org] > > Sent: 25 June 2014 07:39 > > To: Andrew Rae > > Cc: Gene Cumm; syslinux at zytor.com > > Subject: Re: [syslinux] testing out 6.03 network booting... > > > > Andrew, could you try out syslinux-6.03-pre18? Peter pushed the release > > button yesterday and -pre18 contains my change. It would be good to > > know whether it improves things for you. > > Matt Fleming, Intel Open Source Technology Center > > Hi Matt, > Unfortunately I still fail to boot either EFI 64 or EFI 32 bit files past the -pre11/13 era. > > 6.03-pre18 BIOS: lpxelinux.0: boots okay TFTP kernels. > 6.03-pre18 BIOS: lpxelinux.0: boots okay TFTP & HTTP kernels. > 6.03-pre18 EFI32: fails - downloads /pxelinux.e32, /ldlinux.e32, tries /syslinux/ldlinux.e32, tries /boot/isolinux/ldlinux.e32, tries /syslinux/ldlinux.e32, goes back to /ldlinux.e32 - then boot fails > 6.03-pre18 EFI64: fails - exactly same as per EFI32 above but all .e64 instead of .e32 > > Cheers! > AndrewIs there any chance that (at least part of) the behavior depends on the (official) build? For instance, the binaries from ArchLinux currently corresponding to 6.03-pre17 can be downloaded from: https://www.archlinux.org/packages/core/x86_64/syslinux/download/ https://www.archlinux.org/packages/core/i686/syslinux/download/ Also the binaries from Debian Unstable, currently at 6.03-pre17 too, could be downloaded and tested. To be clear, I don't mean that these distros need to be installed, just to download the packages and repeat the test(s) using their respective binaries. Perhaps another possible source of the problem could be related to (the) gnu-efi (submodule). I don't know whether these tests would be worth. If the resulting behavior with these other binaries is the same as with the upstream binaries already tested (in particular comparing with 6.03-pre17), then we would know nothing new. But if the behavior changes... I am also wondering why Gene had a positive result in EFI64 with the same patch from Matt, but Andrew's test failed. Regards, Ady.
>From: Syslinux [syslinux-bounces at zytor.com] On Behalf Of Ady [ady-sf at hotmail.com] >Sent: 25 June 2014 19:32 >To: syslinux at zytor.com >Subject: Re: [syslinux] testing out 6.03 network booting... > >> > From: Matt Fleming [matt at console-pimps.org] >> > Sent: 25 June 2014 07:39 >> > To: Andrew Rae >> > Cc: Gene Cumm; syslinux at zytor.com >> > Subject: Re: [syslinux] testing out 6.03 network booting... >> > >> > Andrew, could you try out syslinux-6.03-pre18? Peter pushed the release >> > button yesterday and -pre18 contains my change. It would be good to >> > know whether it improves things for you. >> > Matt Fleming, Intel Open Source Technology Center >> >> Hi Matt, >> Unfortunately I still fail to boot either EFI 64 or EFI 32 bit files past the -pre11/13 era. >> >> 6.03-pre18 BIOS: lpxelinux.0: boots okay TFTP kernels. >> 6.03-pre18 BIOS: lpxelinux.0: boots okay TFTP & HTTP kernels. >> 6.03-pre18 EFI32: fails - downloads /pxelinux.e32, /ldlinux.e32, tries /syslinux/ldlinux.e32, tries /boot/isolinux/ldlinux.e32, tries /syslinux/ldlinux.e32, goes back to /ldlinux.e32 - then boot fails >> 6.03-pre18 EFI64: fails - exactly same as per EFI32 above but all .e64 instead of .e32 >> >> Cheers! >> Andrew > >Is there any chance that (at least part of) the behavior depends on >the (official) build? > >For instance, the binaries from ArchLinux currently corresponding to >6.03-pre17 can be downloaded from: > > https://www.archlinux.org/packages/core/x86_64/syslinux/download/I tried the version at this page, with the same failure on EFI32 and EFI64 as with kernel.org versions. BIOS (lpxelinux.0) still worked OK.> https://www.archlinux.org/packages/core/i686/syslinux/download/I tried this one too, which contains just EFI32 & BIOS. EFI32 fails in same was as x86_64 version.>Also the binaries from Debian Unstable, currently at 6.03-pre17 too, >could be downloaded and tested. > >To be clear, I don't mean that these distros need to be installed, >just to download the packages and repeat the test(s) using their >respective binaries. > >Perhaps another possible source of the problem could be related to >(the) gnu-efi (submodule). > >I don't know whether these tests would be worth. If the resulting >behavior with these other binaries is the same as with the upstream >binaries already tested (in particular comparing with 6.03-pre17), >then we would know nothing new. But if the behavior changes... > >I am also wondering why Gene had a positive result in EFI64 with the >same patch from Matt, but Andrew's test failed.No idea, I've triple checked I'm not doing something stupid when copying the files to tftp folders etc... and I scripted it so can flit between versions easily - back to pre11 and booted ok....>Regards, >Ady.If I can, I'll try on a physical EFI machine tomorrow (instead of VMWare WS). Wouldn't the two versions I just tried need the patch adding ?? (As mentioned - I failed to figure out how to build from source...!) Hope that helps at all, Andrew
On Thu, Jun 26, 2014 at 4:41 PM, Andrew Rae <A.M.Rae at leeds.ac.uk> wrote:>>I am also wondering why Gene had a positive result in EFI64 with the >>same patch from Matt, but Andrew's test failed. > > No idea, I've triple checked I'm not doing something stupid when copying the files to tftp folders etc... and I scripted it so can flit between versions easily - back to pre11 and booted ok....Probably the same idea why 6.03-pre11 EFI32 fails in my EFI VM. I'm using VMware Workstation 10.0.2, previously a VMX-10 VM and now also a VMX-9 VM, both with e1000 vNICs -- -Gene
On Thu, 26 Jun, at 09:41:38PM, Andrew Rae wrote:> > If I can, I'll try on a physical EFI machine tomorrow (instead of > VMWare WS).Hey Andrew, Sorry for the delay. Could you provide the packet dumps of the network traffic involved in the failed EFI boot? Hopefully that will allow us to track down exactly what's going wrong. -- Matt Fleming, Intel Open Source Technology Center