On 14.07.2015 01:31, Gene Cumm wrote:> On Mon, Jul 13, 2015 at 3:24 PM, poma <pomidorabelisima at gmail.com> wrote: >> On 13.07.2015 18:48, Gene Cumm wrote: > >>> It stops after printing the access type "ETCD" and before printing the >>> remainder of the banner. I presume this is not a hybridized image? >>> >> >> >> I guess, actually it does support hybrid booting. > > So it's hybridized. Could you test without the hybridization? Could > you post or privately email me core/isolinux.bin as built in the > source tree and before being used in an image and its ldlinux.c32? >Vulgaris "make bios" & "rpmbuild" builds. isolinux ??? syslinux-5186539-dollar-sign-then-the-caret-patched ? ??? bios ? ??? com32 ? ? ??? elflink ? ? ??? ldlinux ? ? ??? ldlinux.c32 ? ??? core ? ??? isolinux.bin ??? syslinux-5186539-origin ? ??? bios ? ??? com32 ? ? ??? elflink ? ? ??? ldlinux ? ? ??? ldlinux.c32 ? ??? core ? ??? isolinux.bin ??? syslinux-nonlinux-6.04-3.git20150627.fc23.noarch ? ??? usr ? ??? share ? ??? syslinux ? ??? isolinux.bin ? ??? ldlinux.c32 ??? syslinux-nonlinux-6.04-4.git20150627.fc23.noarch-dollar-sign-then-the-caret-patched ??? usr ??? share ??? syslinux ??? isolinux.bin ??? ldlinux.c32 20 directories, 8 files http://goo.gl/Gm4ffO ISOLINUX/isolinux.tar -------------- next part -------------- A non-text attachment was scrubbed... Name: isolinux-non-isohybrid.png Type: image/png Size: 2870 bytes Desc: not available URL: <http://www.zytor.com/pipermail/syslinux/attachments/20150714/2102df7e/attachment.png>
On Tue, Jul 14, 2015 at 3:23 AM, poma <pomidorabelisima at gmail.com> wrote:> On 14.07.2015 01:31, Gene Cumm wrote: >> On Mon, Jul 13, 2015 at 3:24 PM, poma <pomidorabelisima at gmail.com> wrote: >>> On 13.07.2015 18:48, Gene Cumm wrote: >> >>>> It stops after printing the access type "ETCD" and before printing the >>>> remainder of the banner. I presume this is not a hybridized image? >>>> >>> >>> >>> I guess, actually it does support hybrid booting. >> >> So it's hybridized. Could you test without the hybridization? Could >> you post or privately email me core/isolinux.bin as built in the >> source tree and before being used in an image and its ldlinux.c32? >> > > > Vulgaris "make bios" & "rpmbuild" builds.Although it appears unaffected by hybridization, it's still something to check. I started reading over my recent build of bios/core/isolinux.lst (which should reflect how the ASM of isolinux.bin looks in RAM), I found there's a bunch of protected mode calls before this. isolinux.asm:1119:init.inc:32 "pm_call pm_decompress" should be the first. A similar push/mov/call/pop should help. -- -Gene
Hi, i wrote:> > Do we have indications that different versions of gcc5 > > cause different behavior on the same build and boot machines ?Gene Cumm wrote:> William Kennington's report is definitely different than poma's > original report. [...] > William's report indicated "5.2-rc".Yes. If i got it right then he tried two gcc5 versions which both did not produce bootable binaries. My question was whether in the same particular production environment the produced binaries from different gcc5 versions show significant differences when booting. (This could indicate active work in the compiler region which would cause the troubles.) But meanwhile the suspicion seems to have wandered to assembler and linker rather than C compiler.> Although it appears unaffected by hybridization, it's still something to > check.isohybrid.c and isohybrid[.pl] do not alter the content of isolinux.bin. They only determine its block address (from El Torito catalog) which they write together with the MBR code to the start of the ISO 9660 filesystem (the System Area). Booting via BIOS from a (virtual) DVD drive is not supposed to even notice the presence of MBR. So, if ever, it would be the suitability for isohybrid, not the effective application of isohybrid, which makes the difference for the compiler, or assembler, or linker. Have a nice day :) Thomas