Syslinux - Dec 2017 - Compilation of syslinux against musl libc

Hello, 

I wanted to ask if there's a reason efi/wrapper.{h,c} use __uint*_t
instead of including stdint.h and using the therein defined uint*_t
types? This breaks compilation of the efi blobs on systems with the musl
libc since that doesn't define __uint*_t types.

-- 
Simon Thelen

On Sat, Dec 2, 2017 at 1:32 PM, Simon Thelen via Syslinux
<syslinux at zytor.com> wrote:
> Hello,
>
> I wanted to ask if there's a reason efi/wrapper.{h,c} use __uint*_t
> instead of including stdint.h and using the therein defined uint*_t
> types? This breaks compilation of the efi blobs on systems with the musl
> libc since that doesn't define __uint*_t types.
Good question.  Those two and some XFS stuff are the only instances.
I thought the __uint{8,16,32,64}_t were more glibc specific.

Matt/HPA,any thoughts?

-- 
-Gene

> Hello, 
> 
> I wanted to ask if there's a reason efi/wrapper.{h,c} use __uint*_t
> instead of including stdint.h and using the therein defined uint*_t
> types? This breaks compilation of the efi blobs on systems with the musl
> libc since that doesn't define __uint*_t types.
> 
> -- 
> Simon Thelen
 
FWIW...

As a simple quick basic test, I took the efi32 binaries (syslinux.efi 
and ldlinux.e32, version 6.04-pre1) from alpinelinux (which uses musl) 
and put them in a FAT 1440KiB floppy image using the default UEFI path.

While booting a VM with such floppy image in EFI mode, I arrive to the 
syslinux boot prompt.

I have not attempted to execute any kernel, nor any c32 module, but the 
most basic functionality (i.e. booting to the boot prompt) worked, so I 
assume that the binaries were successfully built (i.e. if there was 
some building issue, it was not completely catastrophic).

Regards,
Ady.

On 17-12-03 at 17:17, Ady Ady via Syslinux wrote:
> FWIW...
> 
> As a simple quick basic test, I took the efi32 binaries (syslinux.efi 
> and ldlinux.e32, version 6.04-pre1) from alpinelinux (which uses musl) 
> and put them in a FAT 1440KiB floppy image using the default UEFI path.
> 
> While booting a VM with such floppy image in EFI mode, I arrive to the 
> syslinux boot prompt.
> 
> I have not attempted to execute any kernel, nor any c32 module, but the 
> most basic functionality (i.e. booting to the boot prompt) worked, so I 
> assume that the binaries were successfully built (i.e. if there was 
> some building issue, it was not completely catastrophic).
I actually checked this and Alpine linux uses the pre-built binary blobs
that lie in the syslinux tarfile [1]. It only builds the installer
target and then installs the files without building the subdirs target
thereby bypassing the build failure in the efi subdirectory.

[1]: https://git.alpinelinux.org/cgit/aports/tree/main/syslinux/APKBUILD

-- 
Simon Thelen