asm-generic/unistd.h: handle symbol prefixes in cond_syscall

Some architectures have symbol prefixes and set CONFIG_SYMBOL_PREFIX, but this wasn't taken into account by the generic cond_syscall. It's easy enough to fix in a generic fashion, so add the symbol prefix to symbol names in cond_syscall when CONFIG_SYMBOL_PREFIX is set.

Signed-off-by James Hogan <[email protected]>

---
 include/asm-generic/unistd.h |    9 ++++++++-
 1 files changed, 8 insertions(+), 1 deletions(-)

include/asm-generic/unistd.h +0 -0

index 2292d1a..c9a5ba4 100644
@@ -924,7 +924,14 @@ __SYSCALL(__NR_fork, sys_ni_syscall)
  * but it doesn't work on all toolchains, so we just do it by hand
  */
 #ifndef cond_syscall

Quoted 18 lines · James Hogan

Some architectures have symbol prefixes and set CONFIG_SYMBOL_PREFIX, but this wasn't taken into account by the generic cond_syscall. It's easy enough to fix in a generic fashion, so add the symbol prefix to symbol names in cond_syscall when CONFIG_SYMBOL_PREFIX is set.

Signed-off-by James Hogan <[email protected]>

---
 include/asm-generic/unistd.h |    9 ++++++++-
 1 files changed, 8 insertions(+), 1 deletions(-)

include/asm-generic/unistd.h +0 -0

index 2292d1a..c9a5ba4 100644
@@ -924,7 +924,14 @@ __SYSCALL(__NR_fork, sys_ni_syscall)
  * but it doesn't work on all toolchains, so we just do it by hand
  */
 #ifndef cond_syscall

include/asm-generic/unistd.h __SYSCALL(__NR_fork, sys_ni_syscall) · line 935

Quoted 10 lines

include/asm-generic/unistd.h +8 -1

-#define cond_syscall(x) asm(".weak\t" #x "\n\t.set\t" #x ",sys_ni_syscall")
+#ifdef CONFIG_SYMBOL_PREFIX
+#define __SYMBOL_PREFIX CONFIG_SYMBOL_PREFIX
+#else
+#define __SYMBOL_PREFIX
+#endif
+#define cond_syscall(x) asm(".weak\t" __SYMBOL_PREFIX #x "\n\t" \
+                           ".set\t" __SYMBOL_PREFIX #x "," \
+                           __SYMBOL_PREFIX "sys_ni_syscall")
 #endif

Our trend is to move away from arch specific Kconfig symbols and __ARCH_HAS_* macros towards just defining whatever you need in the architecture as an override for the generic definition.

Just provide your own unistd.h that does

#define cond_syscall(x) asm(".weak\t." #x "\n\t.set\t." #x ",.sys_ni_syscall") #include <asm-generic/unistd.h>

BTW, are you planning to submit arch/metag for inclusion? I've looked at the code recently and it appears that you are on the right track overall, and it shouldn't be too hard to get to the same quality level as arch/openrisc.

I've been looking at your initial comments for other architecture submissions and they've been very helpful, but if you do have any arch/metag specific suggestions I'd certainly welcome them.

Quoted 3 lines · James Hogan

I've been looking at your initial comments for other architecture submissions and they've been very helpful, but if you do have any arch/metag specific suggestions I'd certainly welcome them.

I've had only a brief look. In general I recommend doing whatever openrisc does, they are currently the best implementation we have.

In order to get the code merged, you will need to move probing of on-chip buses to device trees rather than hardcoding the platform devices, that is probably the largest amount of work that is still needed unless you've done that already in a later version.

Do you have a git tree or patch with the latest source code? the only copy I could find is a 2.6.37 kernel inside of a huge tarball.

Quoted 23 lines · James Hogan

Some architectures have symbol prefixes and set CONFIG_SYMBOL_PREFIX, but this wasn't taken into account by the generic cond_syscall. It's easy enough to fix in a generic fashion, so add the symbol prefix to symbol names in cond_syscall when CONFIG_SYMBOL_PREFIX is set.

 include/asm-generic/unistd.h |    9 ++++++++-
 1 files changed, 8 insertions(+), 1 deletions(-)

include/asm-generic/unistd.h +0 -1

 
 #ifndef cond_syscall
-#define cond_syscall(x) asm(".weak\t" #x "\n\t.set\t" #x

",sys_ni_syscall")

include/asm-generic/unistd.h +8 -0

+#ifdef CONFIG_SYMBOL_PREFIX
+#define __SYMBOL_PREFIX CONFIG_SYMBOL_PREFIX
+#else
+#define __SYMBOL_PREFIX
+#endif
+#define cond_syscall(x) asm(".weak\t" __SYMBOL_PREFIX #x "\n\t" \
+                           ".set\t" __SYMBOL_PREFIX #x "," \
+                           __SYMBOL_PREFIX "sys_ni_syscall")

Quoted 1 line

#endif

Our trend is to move away from arch specific Kconfig symbols and __ARCH_HAS_* macros towards just defining whatever you need in the architecture as an override for the generic definition.

Just provide your own unistd.h that does

Quoted 3 lines

Our trend is to move away from arch specific Kconfig symbols and __ARCH_HAS_* macros towards just defining whatever you need in the architecture as an override for the generic definition.

i don't see how __ARCH_HAS_xxx would help here. the symbol prefix is a string, not a bool value. it's also already used by linux/export.h, asm- generic/vmlinux.lds.h, and module code in scripts/.

Quoted 1 line

Just provide your own unistd.h that does

the point of asm-generic is so that arches don't have to keep copying & pasting things that they really don't care about. James' proposed patch looks good to me. it might be nice to go even further and add logic to a core header so that CONFIG_SYMBOL_PREFIX is always defined ...

We actually have MODULE_SYMBOL_PREFIX providing the correct string under a name that is not appropriate here. We can probably stick with Jason's patch for now and add a more sophisticated logic if another user of CONFIG_SYMBOL_PREFIX comes up.

So for Jason's patch:

Acked-by Arnd Bergmann <[email protected]>

We could put that into the kernel now, but it's probably sufficient if Jason keeps the patch with his others and submits it at the same time when we get there.

Quoted 7 lines · Arnd Bergmann

We actually have MODULE_SYMBOL_PREFIX providing the correct string under a name that is not appropriate here. We can probably stick with Jason's patch for now and add a more sophisticated logic if another user of CONFIG_SYMBOL_PREFIX comes up.

So for Jason's patch:

Quoted 5 lines

Acked-by Arnd Bergmann <[email protected]>

We could put that into the kernel now, but it's probably sufficient if Jason keeps the patch with his others and submits it at the same time when we get there.

Sorry, I meant s/Jason/James/g

Some architectures have symbol prefixes and set CONFIG_SYMBOL_PREFIX, but this wasn't taken into account by the generic cond_syscall. It's easy enough to fix in a generic fashion, so add the symbol prefix to symbol names in cond_syscall when CONFIG_SYMBOL_PREFIX is set.

Signed-off-by James Hogan <[email protected]>

---
 include/asm-generic/unistd.h |    9 ++++++++-
 1 files changed, 8 insertions(+), 1 deletions(-)

include/asm-generic/unistd.h +0 -0

index 2292d1a..c9a5ba4 100644
@@ -924,7 +924,14 @@ __SYSCALL(__NR_fork, sys_ni_syscall)
  * but it doesn't work on all toolchains, so we just do it by hand
  */
 #ifndef cond_syscall

Some architectures have symbol prefixes and set CONFIG_SYMBOL_PREFIX, but this wasn't taken into account by the generic cond_syscall. It's easy enough to fix in a generic fashion, so add the symbol prefix to symbol names in cond_syscall when CONFIG_SYMBOL_PREFIX is set.