commit 30fb9454ab23 ("selftests/vm: hmm-tests: remove the libhugetlbfs
dependency") created the new check_config.sh file without the execute
bits set. This is a problem because the Makefile runs it with "./", so
now "make" is failing in that directory.
The posted patch [1] does seem to have it as an executable file, so I'm
not sure how this happened. But in any case, make check_config.sh
executable again, so that "make" in selftests/vm works once again.
[1] https://lore.kernel.org/linux-doc/[email protected]/
Fixes: commit 30fb9454ab23 ("selftests/vm: hmm-tests: remove the libhugetlbfs dependency")
Signed-off-by: John Hubbard <[email protected]>
---
Hi,
This applies to today's 20201001) linux-next. I forgot if this needs to
be a separate patch or if it can be fixed up in Andrew's tree.
thanks,
John Hubbard
NVIDIA
tools/testing/selftests/vm/check_config.sh | 0
1 file changed, 0 insertions(+), 0 deletions(-)
mode change 100644 => 100755 tools/testing/selftests/vm/check_config.sh
diff --git a/tools/testing/selftests/vm/check_config.sh b/tools/testing/selftests/vm/check_config.sh
old mode 100644
new mode 100755
--
2.28.0
On 10/1/20 11:48 PM, John Hubbard wrote:
> commit 30fb9454ab23 ("selftests/vm: hmm-tests: remove the libhugetlbfs
> dependency") created the new check_config.sh file without the execute
> bits set. This is a problem because the Makefile runs it with "./", so
> now "make" is failing in that directory.
>
> The posted patch [1] does seem to have it as an executable file, so I'm
> not sure how this happened. But in any case, make check_config.sh
> executable again, so that "make" in selftests/vm works once again.
>
> [1] https://lore.kernel.org/linux-doc/[email protected]/
>
> Fixes: commit 30fb9454ab23 ("selftests/vm: hmm-tests: remove the libhugetlbfs dependency")
> Signed-off-by: John Hubbard <[email protected]>
> ---
>
> Hi,
>
> This applies to today's 20201001) linux-next. I forgot if this needs to
> be a separate patch or if it can be fixed up in Andrew's tree.
>
hmm, in fact I think something similar also happened to run_vmtest.sh, in the same
directory. Please let me know if a patch is needed there, or if you can just fix
up this one or the tree or something?
> thanks,
> John Hubbard
> NVIDIA
>
>
> tools/testing/selftests/vm/check_config.sh | 0
> 1 file changed, 0 insertions(+), 0 deletions(-)
> mode change 100644 => 100755 tools/testing/selftests/vm/check_config.sh
>
> diff --git a/tools/testing/selftests/vm/check_config.sh b/tools/testing/selftests/vm/check_config.sh
> old mode 100644
> new mode 100755
>
thanks,
--
John Hubbard
NVIDIA
Hi John,
On Thu, 1 Oct 2020 23:48:11 -0700 John Hubbard <[email protected]> wrote:
>
> commit 30fb9454ab23 ("selftests/vm: hmm-tests: remove the libhugetlbfs
> dependency") created the new check_config.sh file without the execute
> bits set. This is a problem because the Makefile runs it with "./", so
> now "make" is failing in that directory.
>
> The posted patch [1] does seem to have it as an executable file, so I'm
> not sure how this happened. But in any case, make check_config.sh
> executable again, so that "make" in selftests/vm works once again.
The correct fix is to make sure that the Makefile runs it explicitly
with a shell because is such a patch is applied with patch(1), the
execute bit will not be preserved.
See e.g. in the top level Makefile:
$(CONFIG_SHELL) $(srctree)/scripts/gen_autoksyms.sh $@
--
Cheers,
Stephen Rothwell
On 10/2/20 12:59 AM, Stephen Rothwell wrote:
> Hi John,
>
> On Thu, 1 Oct 2020 23:48:11 -0700 John Hubbard <[email protected]> wrote:
>>
>> commit 30fb9454ab23 ("selftests/vm: hmm-tests: remove the libhugetlbfs
>> dependency") created the new check_config.sh file without the execute
>> bits set. This is a problem because the Makefile runs it with "./", so
>> now "make" is failing in that directory.
>>
>> The posted patch [1] does seem to have it as an executable file, so I'm
>> not sure how this happened. But in any case, make check_config.sh
>> executable again, so that "make" in selftests/vm works once again.
>
> The correct fix is to make sure that the Makefile runs it explicitly
> with a shell because is such a patch is applied with patch(1), the
> execute bit will not be preserved.
>
> See e.g. in the top level Makefile:
>
> $(CONFIG_SHELL) $(srctree)/scripts/gen_autoksyms.sh $@
>
OK...we should also fix it anyway, seeing as how the run_vmtest.sh also needs
the execute bit re-set, yes?
So I guess I should send a patch that does:
* chmod +x on both run_vmtest.sh and check_config.sh, plus
* invoke check_config.sh via $(CONFIG_SHELL) from the Makefile.
Does that sound about right?
thanks,
--
John Hubbard
NVIDIA
On 10/2/20 1:10 AM, John Hubbard wrote:
> On 10/2/20 12:59 AM, Stephen Rothwell wrote:
>> Hi John,
>>
>> On Thu, 1 Oct 2020 23:48:11 -0700 John Hubbard <[email protected]> wrote:
>>>
>>> commit 30fb9454ab23 ("selftests/vm: hmm-tests: remove the libhugetlbfs
>>> dependency") created the new check_config.sh file without the execute
>>> bits set. This is a problem because the Makefile runs it with "./", so
>>> now "make" is failing in that directory.
>>>
>>> The posted patch [1] does seem to have it as an executable file, so I'm
>>> not sure how this happened. But in any case, make check_config.sh
>>> executable again, so that "make" in selftests/vm works once again.
>>
>> The correct fix is to make sure that the Makefile runs it explicitly
>> with a shell because is such a patch is applied with patch(1), the
>> execute bit will not be preserved.
>>
>> See e.g. in the top level Makefile:
>>
>> ????$(CONFIG_SHELL) $(srctree)/scripts/gen_autoksyms.sh $@
>>
>
> OK...we should also fix it anyway, seeing as how the run_vmtest.sh also needs
> the execute bit re-set, yes?
>
...and I should add that I'm stuck in a non-Kbuild system, here. So
$(CONFIG_SHELL) is not available, sigh.
> So I guess I should send a patch that does:
>
> * chmod +x on both run_vmtest.sh and check_config.sh, plus
>
> * invoke check_config.sh via $(CONFIG_SHELL) from the Makefile.
>
> Does that sound about right?
>
> thanks,
thanks,
--
John Hubbard
NVIDIA