Commit 4e21565b7fd4 ("asm-generic: add kexec_file_load system
call to unistd.h") added the system call to the generic header
but not to the perf copy resulting a compile failure on aarch64
When running mksyscalltbl
BUILDSTDERR: <stdin>: In function 'main':
BUILDSTDERR: <stdin>:273:44: error: '__NR_kexec_file_load' undeclared (first use in this function)
BUILDSTDERR: <stdin>:273:44: note: each undeclared identifier is reported only once for each function it appears in
Fix this by syncing up.
Fixes: 4e21565b7fd4 ("asm-generic: add kexec_file_load system call to unistd.h")
Signed-off-by: Laura Abbott <[email protected]>
---
Found this on Fedora when compiling 5.0-rc1, I hadn't seen a patch
queued yet.
---
tools/include/uapi/asm-generic/unistd.h | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/tools/include/uapi/asm-generic/unistd.h b/tools/include/uapi/asm-generic/unistd.h
index c7f3321fbe43..d90127298f12 100644
--- a/tools/include/uapi/asm-generic/unistd.h
+++ b/tools/include/uapi/asm-generic/unistd.h
@@ -738,9 +738,11 @@ __SYSCALL(__NR_statx, sys_statx)
__SC_COMP(__NR_io_pgetevents, sys_io_pgetevents, compat_sys_io_pgetevents)
#define __NR_rseq 293
__SYSCALL(__NR_rseq, sys_rseq)
+#define __NR_kexec_file_load 294
+__SYSCALL(__NR_kexec_file_load, sys_kexec_file_load)
#undef __NR_syscalls
-#define __NR_syscalls 294
+#define __NR_syscalls 295
/*
* 32 bit systems traditionally used different
--
2.20.1
Em Tue, Jan 08, 2019 at 02:17:58PM -0800, Laura Abbott escreveu:
> Commit 4e21565b7fd4 ("asm-generic: add kexec_file_load system
> call to unistd.h") added the system call to the generic header
> but not to the perf copy resulting a compile failure on aarch64
Humm, that shouldn't happen, i.e. kernel developers don't have to update
anything in tools/ as part of their normal kernel development workflows.
They can if they wish, but are not required, so the build failure was
due to something else or a pre-existing bug where tools/ living stuff
used things outside tools/.
> When running mksyscalltbl
Lemme try this on my Orangi PI zero...
- Arnaldo
> BUILDSTDERR: <stdin>: In function 'main':
> BUILDSTDERR: <stdin>:273:44: error: '__NR_kexec_file_load' undeclared (first use in this function)
> BUILDSTDERR: <stdin>:273:44: note: each undeclared identifier is reported only once for each function it appears in
>
> Fix this by syncing up.
>
> Fixes: 4e21565b7fd4 ("asm-generic: add kexec_file_load system call to unistd.h")
> Signed-off-by: Laura Abbott <[email protected]>
> ---
> Found this on Fedora when compiling 5.0-rc1, I hadn't seen a patch
> queued yet.
> ---
> tools/include/uapi/asm-generic/unistd.h | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/tools/include/uapi/asm-generic/unistd.h b/tools/include/uapi/asm-generic/unistd.h
> index c7f3321fbe43..d90127298f12 100644
> --- a/tools/include/uapi/asm-generic/unistd.h
> +++ b/tools/include/uapi/asm-generic/unistd.h
> @@ -738,9 +738,11 @@ __SYSCALL(__NR_statx, sys_statx)
> __SC_COMP(__NR_io_pgetevents, sys_io_pgetevents, compat_sys_io_pgetevents)
> #define __NR_rseq 293
> __SYSCALL(__NR_rseq, sys_rseq)
> +#define __NR_kexec_file_load 294
> +__SYSCALL(__NR_kexec_file_load, sys_kexec_file_load)
>
> #undef __NR_syscalls
> -#define __NR_syscalls 294
> +#define __NR_syscalls 295
>
> /*
> * 32 bit systems traditionally used different
> --
> 2.20.1
--
- Arnaldo
On 1/9/19 5:15 AM, Arnaldo Carvalho de Melo wrote:
> Em Tue, Jan 08, 2019 at 02:17:58PM -0800, Laura Abbott escreveu:
>> Commit 4e21565b7fd4 ("asm-generic: add kexec_file_load system
>> call to unistd.h") added the system call to the generic header
>> but not to the perf copy resulting a compile failure on aarch64
>
> Humm, that shouldn't happen, i.e. kernel developers don't have to update
> anything in tools/ as part of their normal kernel development workflows.
>
> They can if they wish, but are not required, so the build failure was
> due to something else or a pre-existing bug where tools/ living stuff
> used things outside tools/.
>
I think that's the issue, I couldn't reproduce the issue until I
installed the 5.0-rc1 headers. This is the invocation of the script
(paths trimmed)
/tools/perf/arch/arm64/entry/syscalls//mksyscalltbl' 'gcc' 'gcc' /tools /tools/arch/arm64/include/uapi/asm/unistd.h > arch/arm64/include/generated/asm/syscalls.c
In tools/perf/arch/arm64/entry/syscalls/mksyscalltbl,
$hostcc -I $incpath/include/uapi -o $create_table_exe -x c -
This was the gcc command that was failing.
tools/arch/arm64/include/uapi/asm/unistd.h includes asm-generic/unistd.h
so I think that's what's leaking from the system.
>> When running mksyscalltbl
>
> Lemme try this on my Orangi PI zero...
>
> - Arnaldo
> >> BUILDSTDERR: <stdin>: In function 'main':
>> BUILDSTDERR: <stdin>:273:44: error: '__NR_kexec_file_load' undeclared (first use in this function)
>> BUILDSTDERR: <stdin>:273:44: note: each undeclared identifier is reported only once for each function it appears in
>>
>> Fix this by syncing up.
>>
>> Fixes: 4e21565b7fd4 ("asm-generic: add kexec_file_load system call to unistd.h")
>> Signed-off-by: Laura Abbott <[email protected]>
>> ---
>> Found this on Fedora when compiling 5.0-rc1, I hadn't seen a patch
>> queued yet.
>> ---
>> tools/include/uapi/asm-generic/unistd.h | 4 +++-
>> 1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/tools/include/uapi/asm-generic/unistd.h b/tools/include/uapi/asm-generic/unistd.h
>> index c7f3321fbe43..d90127298f12 100644
>> --- a/tools/include/uapi/asm-generic/unistd.h
>> +++ b/tools/include/uapi/asm-generic/unistd.h
>> @@ -738,9 +738,11 @@ __SYSCALL(__NR_statx, sys_statx)
>> __SC_COMP(__NR_io_pgetevents, sys_io_pgetevents, compat_sys_io_pgetevents)
>> #define __NR_rseq 293
>> __SYSCALL(__NR_rseq, sys_rseq)
>> +#define __NR_kexec_file_load 294
>> +__SYSCALL(__NR_kexec_file_load, sys_kexec_file_load)
>>
>> #undef __NR_syscalls
>> -#define __NR_syscalls 294
>> +#define __NR_syscalls 295
>>
>> /*
>> * 32 bit systems traditionally used different
>> --
>> 2.20.1
>