Gene Cumm wrote:> On Tue, Dec 1, 2015 at 3:02 PM, Erwan Velu via Syslinux > <syslinux at zytor.com> wrote: > > Hi folks, > > > > As per Ady request (thanks for the reminder), I forgot to send the patches > > I run in Mageia. > > > > There is two of them : > > > > One is making errors in parsing modules.alias not fatal. > > > > The other one is adding an option to chain to avoid reading garbage on the > > command line. > > I had some issues with gfxboot and had to implement that. I know it's a > > workaround, I don't really get if that's useful for others. You can forget > > it if you consider it as too weird. > > This almost sounds like a global APPEND was used. Current Syslinux > should ignore the global APPEND for COM32 use but other config parsers > like (vesa)menu.c32 and gfxboot.c32 might not as they don't call by > LABEL but execute plain commands.How can I reproduce this issue? Sebastian
On Thu, Dec 10, 2015 at 5:15 PM, Sebastian Herbszt <herbszt at gmx.de> wrote:> Gene Cumm wrote: >> On Tue, Dec 1, 2015 at 3:02 PM, Erwan Velu via Syslinux >> <syslinux at zytor.com> wrote: >> > Hi folks, >> > >> > As per Ady request (thanks for the reminder), I forgot to send the patches >> > I run in Mageia. >> > >> > There is two of them : >> > >> > One is making errors in parsing modules.alias not fatal. >> > >> > The other one is adding an option to chain to avoid reading garbage on the >> > command line. >> > I had some issues with gfxboot and had to implement that. I know it's a >> > workaround, I don't really get if that's useful for others. You can forget >> > it if you consider it as too weird. >> >> This almost sounds like a global APPEND was used. Current Syslinux >> should ignore the global APPEND for COM32 use but other config parsers >> like (vesa)menu.c32 and gfxboot.c32 might not as they don't call by >> LABEL but execute plain commands. > > How can I reproduce this issue? > > Sebastianhttp://rpmfind.net/linux/mageia/distrib/cauldron/i586/isolinux/isolinux.cfg and http://rpmfind.net/linux/mageia/distrib/cauldron/x86_64/isolinux/isolinux.cfg ( http://distro.ibiblio.org/mageia/distrib/cauldron/ appears to be another mirror) look clean from anything that should resemble a global APPEND. Using the files from the above directory, I see "automatic=method:cdrom" appended as an extra argument. Trimming unnecessary files, I can get it to an ISO under 1 MiB (trim hdt.c32, memtest, modules.*, *.ids, *.hlp ). -- -Gene
> On Thu, Dec 10, 2015 at 5:15 PM, Sebastian Herbszt <herbszt at gmx.de> wrote: > > Gene Cumm wrote: > >> On Tue, Dec 1, 2015 at 3:02 PM, Erwan Velu via Syslinux > >> <syslinux at zytor.com> wrote: > >> > Hi folks, > >> > > >> > As per Ady request (thanks for the reminder), I forgot to send the patches > >> > I run in Mageia. > >> > > >> > There is two of them : > >> > > >> > One is making errors in parsing modules.alias not fatal. > >> > > >> > The other one is adding an option to chain to avoid reading garbage on the > >> > command line. > >> > I had some issues with gfxboot and had to implement that. I know it's a > >> > workaround, I don't really get if that's useful for others. You can forget > >> > it if you consider it as too weird. > >> > >> This almost sounds like a global APPEND was used. Current Syslinux > >> should ignore the global APPEND for COM32 use but other config parsers > >> like (vesa)menu.c32 and gfxboot.c32 might not as they don't call by > >> LABEL but execute plain commands. > > > > How can I reproduce this issue? > > > > Sebastian > > http://rpmfind.net/linux/mageia/distrib/cauldron/i586/isolinux/isolinux.cfg > and http://rpmfind.net/linux/mageia/distrib/cauldron/x86_64/isolinux/isolinux.cfg > ( http://distro.ibiblio.org/mageia/distrib/cauldron/ appears to be > another mirror) look clean from anything that should resemble a global > APPEND. > > Using the files from the above directory, I see > "automatic=method:cdrom" appended as an extra argument. Trimming > unnecessary files, I can get it to an ISO under 1 MiB (trim hdt.c32, > memtest, modules.*, *.ids, *.hlp ). > > -- > -GeneFWIW, kernel.org also has mirror(s) for mageia (including these directories / files). IIRC, the issue is not related to any global APPEND (mageia does not use it), but to the interaction between gfxboot.c32 and chain.c32. To replicate, see "gfxboot.cfg" and "isolinux.cfg" for mageia (in the aforementioned directories) Relevant lines in these 2 cfg files: ["gfxboot.cfg"] ; main menu items where user can't add boot options nobootoptions=harddisk,memtest ; main menu items that are not passed an 'install' parameter noinstallopt=harddisk,firmware,memtest [/"gfxboot.cfg"] ["isolinux.cfg"] ui gfxboot.c32 bootlogo label harddisk com32 chain.c32 hd0 exit [/"isolinux.cfg"] As usual, using 3 or more menu entries is recommended so to avoid the minor "quirks" when using less than that amount. Please don't forget to use unique LABELs without special characters (such as spaces). Alternatives for potential testing, either with the original Syslinux 6.03, or with later development code before the introduction of the "exit" hack: ["isolinux.cfg"] ui gfxboot.c32 bootlogo label harddisk com32 chain.c32 hd0 # no "exit" [/"isolinux.cfg"] ["isolinux.cfg"] ui gfxboot.c32 bootlogo label harddisk com32 chain.c32 append hd0 exit # using "append" [/"isolinux.cfg"] ["isolinux.cfg"] ui gfxboot.c32 bootlogo label harddisk com32 chain.c32 append hd0 # using "append" but no "exit" [/"isolinux.cfg"] ["isolinux.cfg"] ui gfxboot.c32 bootlogo label harddisk com32 chain.c32 # using no parameter at all, thus using the default [/"isolinux.cfg"] Additional alternative testing:>From the gfxboot screen, exit to the "normal" Syslinux boot prompt and:_ type in "harddisk" (thus, using the label from isolinux.cfg); and/or, _ type in the command, "chain.c32 hd0". The last 2 alternative tests should allow to get the expected result without the interference of gfxboot. This is usually not what is wanted for final users to use; this is just a workaround, and a method to test the interaction with gfxboot. Regards, Ady.> _______________________________________________ > Syslinux mailing list > Submissions to Syslinux at zytor.com > Unsubscribe or set options at: > http://www.zytor.com/mailman/listinfo/syslinux >
Gene Cumm wrote:> On Thu, Dec 10, 2015 at 5:15 PM, Sebastian Herbszt <herbszt at gmx.de> wrote: > > Gene Cumm wrote: > >> On Tue, Dec 1, 2015 at 3:02 PM, Erwan Velu via Syslinux > >> <syslinux at zytor.com> wrote: > >> > Hi folks, > >> > > >> > As per Ady request (thanks for the reminder), I forgot to send the patches > >> > I run in Mageia. > >> > > >> > There is two of them : > >> > > >> > One is making errors in parsing modules.alias not fatal. > >> > > >> > The other one is adding an option to chain to avoid reading garbage on the > >> > command line. > >> > I had some issues with gfxboot and had to implement that. I know it's a > >> > workaround, I don't really get if that's useful for others. You can forget > >> > it if you consider it as too weird. > >> > >> This almost sounds like a global APPEND was used. Current Syslinux > >> should ignore the global APPEND for COM32 use but other config parsers > >> like (vesa)menu.c32 and gfxboot.c32 might not as they don't call by > >> LABEL but execute plain commands. > > > > How can I reproduce this issue? > > > > Sebastian > > http://rpmfind.net/linux/mageia/distrib/cauldron/i586/isolinux/isolinux.cfg > and http://rpmfind.net/linux/mageia/distrib/cauldron/x86_64/isolinux/isolinux.cfg > ( http://distro.ibiblio.org/mageia/distrib/cauldron/ appears to be > another mirror) look clean from anything that should resemble a global > APPEND. > > Using the files from the above directory, I see > "automatic=method:cdrom" appended as an extra argument. Trimming > unnecessary files, I can get it to an ISO under 1 MiB (trim hdt.c32, > memtest, modules.*, *.ids, *.hlp ).I downloaded all files, created an iso and noticed it's crap. gfxboot starts but instead of back.jpg I just get a black background. I didn't check the pre-created isos but the bootlogo from the above mentioned isolinux directory is incomplete and does not contain all required files. So I grabbed mageia-gfxboot-theme-4.5.6.9-1.mga6.src.rpm, corrected IN_CPIO_INSTALL in the Makefile and created a proper bootlogo file with size of ~953k instead of ~337k. With that file I get a proper background and can't reproduce the original issue. Sebastian