similar to: [PATCH 0/2] Do not use the "red zone" on EFI

Hi Ady, Would it work if we removed "ifdef EFI_BUILD" condition and just add -mno-red-zone for all x86_64 builds? If not, do you have any ideas how to pass this flag? This could work, because the patch is adding the -mno-red-zone flag only for x86_64 builds, which are only used in the form of the efi64 target. The efi32 and bios targets are both 32-bit. BTW. I also tried 6.04-pre1 and

From: Sylvain Gault <sylvain.gault at gmail.com> These few patches contain a few improvement about the Makefiles for EFI. Mainly, to rebuild the files when needed, and only when needed. The three shell scripts efi/{check,build,clean}-gnu-efi.sh disappeared and are now integrated as makefile recipes. You'll notice an argument ARFLAGS=rvU to the recursive make calls to gnu-efi. This is

Hello syslinux maintainers, I came across the issue of syslinux crashing with some EFIs in the UEFI boot mode. Upon closer investigation it turned out that most object files which go into syslinux efi64 are compiled with the so-called red zone. Some EFIs follow Windows ABI more strictly and cause syslinux to crash due to stack overwrite. Looking at mk/efi.mk there was a previous attempt to fix

On Nov 28, 2015 2:51 AM, "Ady via Syslinux" <syslinux at zytor.com> wrote: > > From: Sylvain Gault <sylvain.gault at gmail.com> > > > > The System V ABI for x86-64 specify that a "red zone" is an area of 128 bytes > > above the current stack frame. This area can be used by a called function in > > order to avoid the overhead of

On Sat, 22 Jun, at 05:24:21PM, Ferenc Wagner wrote: > Matt Fleming <matt at console-pimps.org> writes: > > > Please do test out the release and report any regressions. > > Unfortunately, I can't make it under Debian wheezy + experimental > gnu-efi (ie. 3.0u+debian-1): > > $ make installer [...] > make[3]: *** No rule to make target

On Mon, Sep 14, 2015 at 05:50:58AM +0200, celelibi--- via Syslinux wrote: > index a705440..5ef6702 100644 > --- a/mk/efi.mk > +++ b/mk/efi.mk > @@ -10,6 +10,7 @@ core = $(topdir)/core > GCCOPT := $(call gcc_ok,-fno-stack-protector,) > EFIINC = $(objdir)/include/efi > LIBDIR = $(objdir)/lib > +EFIDIR = $(topdir)/gnu-efi/gnu-efi-3.0 Would it make sense to add AROPT =

When we add -ffreestanding the compiler won't get to inline this any more. Signed-off-by: David Woodhouse <dwmw2 at infradead.org> --- com32/lib/math/fabs.S | 15 +++++++++++++++ mk/lib.mk | 2 +- 2 files changed, 16 insertions(+), 1 deletion(-) create mode 100644 com32/lib/math/fabs.S diff --git a/com32/lib/math/fabs.S b/com32/lib/math/fabs.S new file mode 100644 index

2015-11-28 8:47 UTC+01:00, Ady via Syslinux <syslinux at zytor.com>: > >> From: Sylvain Gault <sylvain.gault at gmail.com> >> >> The System V ABI for x86-64 specify that a "red zone" is an area of 128 >> bytes >> above the current stack frame. This area can be used by a called function >> in >> order to avoid the overhead of

Thank you all. It looks like -mno-red-zone flag is already in master branch since commit 7d70885d, but it seems it is applied to all EFI builds. IIUC, the problem only affects efi64 builds. So here's the modification of my patch to only apply -mno-red-zone flag to efi64 target. The patch is against master. I verified that it builds correctly in master. Please feel free to accept or reject

From: Sylvain Gault <sylvain.gault at gmail.com> I found something suspicious while hunting another bug a while ago. The conditions for that bug to occur seems quite hard to meet, but it's still code quality improvement. See the commit message for details. Sylvain Gault (1): efi: Suspicious size reduction in emalloc efi/main.c | 4 +--- 1 file changed, 1 insertion(+), 3

From: Sylvain Gault <sylvain.gault at gmail.com> I don't know if I should merge those trivial warning fix into one commit. I can reformat it as requested. Those are a few warning fixes for the efi part. The code involved has mainly been introduced in recent commits. Sylvain Gault (3): efi: fix warnings about argument types efi: fix pointer-type mismatch assigment warning efi: fix

From: Matt Fleming <matt.fleming at linux.intel.com> This series includes a patch (PATCH 1/6) that I sent previously but I thought it was worth sending it again since the rest of the series depends on it, and it also gives a bit of context. These cleanups make it simpler to do the big switchover to ELF modules on the elflink branch because the libraries in $LIBS are now contained in one

From: Sylvain Gault <sylvain.gault at gmail.com> TL;DR: The section aligment in linker scripts messed-up the memory mapping needed for the compression / decompression to work. The bug with gcc 5 is not trivial, I'll do my best to explain it here. Basically, there are two memory mappings of the code. One in "virtual memory", and one in "load memory". The one in

On 10/12/15 21:04, celelibi--- via Syslinux wrote: > From: Sylvain Gault <sylvain.gault at gmail.com> > > The behavior when running a "CONFIG" command line is to reload > ldlinux.c32 with the new file as argument. This call never return. > > In order to avoid stacking up the calls to start_ldlinux, this patch > introduce a setjmp/longjmp to return to the

From: Sylvain Gault <sylvain.gault at gmail.com> Hello there, I propose 2 patches that fix two possible stack overflows either when running a COM32 module or when loading a new config file. I didn't find a better way to do this than to use the infamous setjmp/longjmp functions to restore the stack to a previous state. This makes the logic a bit more complex, but the behavior is not

From: Sylvain Gault <sylvain.gault at gmail.com> The error-prone shell scripts for building gnu-efi are replaced by a Makefile recipe. This is accompanied with a small update of gnu-efi which, despite not strongly mandatory, avoid recompiling gnu-efi and all the efi subtree on each invocation of make. Signed-off-by: Sylvain Gault <sylvain.gault at gmail.com> ---

From: Sylvain Gault <sylvain.gault at gmail.com> The assembly instruction lgdt take a segment limit that is one less than the actual size, so that base+limit points to the last byte. Signed-off-by: Sylvain Gault <sylvain.gault at gmail.com> --- efi/main.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/efi/main.c b/efi/main.c index 94878f9..bdf9353 100644

From: Sylvain Gault <sylvain.gault at gmail.com> Missing */ at the end of a comment. Signed-off-by: Sylvain Gault <sylvain.gault at gmail.com> --- efi/x86_64/linux.S | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/efi/x86_64/linux.S b/efi/x86_64/linux.S index 972c0b2..29dde94 100644 --- a/efi/x86_64/linux.S +++ b/efi/x86_64/linux.S @@ -27,7 +27,7 @@ kernel_jump:

From: Sylvain Gault <sylvain.gault at gmail.com> On some architecture, including my hardware, the function ExitBootServices may need to be called twice in order to successfully exit the boot services. As stated by the UEFI spec, the first call to ExitBootServices may perform a partial shutdown of the services. It seems that during this partial shutdown, the memory map can be modified, thus

From: Sylvain Gault <sylvain.gault at gmail.com> The command-line parsing used to return a pointer to a local array. The code used to work by chance, but now, gcc 5 is able to detect it and return a NULL pointer instead. The buffer is now marked static. This shouldn't be a problem as only one command line can be read at a time. Signed-off-by: Sylvain Gault <sylvain.gault at