search for: syslinux_filesystem

PXELINUX has stopped recognizing the keeppxe option on the kernel command line. Here is a patch to make it work again. With COM32, it is no longer possible to use preprocessor directives to determine the SYSLINUX variant. The code inside the #if will never be compiled. So, I changed it to use syslinux_filesystem() to determine the variant. Also, I moved the relevant code from kernel.c:new_linux_kernel() to load_linux.c:bios_boot_linux() because there is no convenient way in new_linux_kernel() to control the boot flags value. In basic testing, keeppxe seems to work with the patch applied. --- syslinu...

> Pushed. Thank you for digging so deeply on this. > Gene, For the c32 modules to be able to use the keeppxe option, and considering that the issue in v.5.+ is that 'IS_PXELINUX' is/was dead code, is there any chance that the code in: com32/modules/whichsys.c would help? I mean, as a different way of detecting the Syslinux variant. In fact, the original idea of having the

...d recognizing the keeppxe option on the kernel > command line. Here is a patch to make it work again. > > With COM32, it is no longer possible to use preprocessor directives to > determine the SYSLINUX variant. The code inside the #if will never be > compiled. So, I changed it to use syslinux_filesystem() to determine the > variant. Also, I moved the relevant code from > kernel.c:new_linux_kernel() to load_linux.c:bios_boot_linux() because > there is no convenient way in new_linux_kernel() to control the boot > flags value. > > In basic testing, keeppxe seems to work with the pat...

...er possible to use preprocessor directives to Not COM32 but ldlinux.c32 as a result of splitting the core binary in two with variant-specific code to load another file of common code. > determine the SYSLINUX variant. The code inside the #if will never be > compiled. So, I changed it to use syslinux_filesystem() to determine the > variant. Also, I moved the relevant code from Probably the best choice. > kernel.c:new_linux_kernel() to load_linux.c:bios_boot_linux() because > there is no convenient way in new_linux_kernel() to control the boot > flags value. This is the part that has me ques...

...32/lib/sys/gpxe.c | 4 ++-- 2 files changed, 3 insertions(+), 2 deletions(-) diff --git a/com32/include/syslinux/config.h b/com32/include/syslinux/config.h index 868b0f1..2874665 100644 --- a/com32/include/syslinux/config.h +++ b/com32/include/syslinux/config.h @@ -44,6 +44,7 @@ enum syslinux_filesystem { SYSLINUX_FS_PXELINUX = 0x32, SYSLINUX_FS_ISOLINUX = 0x33, SYSLINUX_FS_EXTLINUX = 0x34, + SYSLINUX_FS_GPXE = 0x46, }; struct syslinux_version { diff --git a/com32/lib/sys/gpxe.c b/com32/lib/sys/gpxe.c index fae03f8..1b01abc 100644 --- a/com32/lib/sys/gpxe.c +++ b/com32/lib/sy...

From: Matt Fleming <matt.fleming at intel.com> Since we can't use __intcall() for EFI, and since we can now have the ELF module code resolve all our symbols at runtime, we should delete as many references to __intcall() as possible and just access the symbols directly. The most interesting patch is the support for weak symbols. We need to be able to reference derivative-specific

...slinux/config.h b/com32/include/syslinux/config.h index 50bd52f..7bdcdd6 100644 --- a/com32/include/syslinux/config.h +++ b/com32/include/syslinux/config.h @@ -181,4 +181,9 @@ static inline const struct syslinux_ipappend_strings return &__syslinux_ipappend_strings; } +static inline enum syslinux_filesystem syslinux_filesystem(void) +{ + return syslinux_derivative_info()->c.filesystem; +} + #endif /* _SYSLINUX_CONFIG_H */ diff --git a/com32/include/syslinux/features.h b/com32/include/syslinux/features.h index 4bebda4..d25d08d 100644 --- a/com32/include/syslinux/features.h +++ b/com32/include/sy...