[PATCH v2] linux: Add mseal syscall support

[Adhemerval Zanella Netto]

On 10/11/25 20:43, Collin Funk wrote:
> Hi Adhemerval,
> 
> Adhemerval Zanella <adhemerval.zanella@linaro.org> writes:
> 
>> +@item ENOSYS
>> +The kernel does not support the @code{mseal} syscall.
>> +
>> +@strong{NB:} The memory sealing changes the lifetime of a mapping, where the
>> +sealing memory could not be unmapped until the process terminates or replaces
>> +the process image through @code{execve} function.  The sealed mappings are
>> +inherited through @code{fork}.
>> +
>> +@end table
>> +@end deftypefun
> 
> Shouldn't the section starting with @strong{NB:} be moved outside of the
> table? Otherwise it looks as if the text is related to the failed call
> with ENOSYS. Can you take a look at the PDF output before and after
> moving outside of the table to see what I mean?

It should, I will move it outside.

> 
>> diff --git a/sysdeps/unix/sysv/linux/tst-mseal-pkey.c b/sysdeps/unix/sysv/linux/tst-mseal-pkey.c
>> new file mode 100644
>> index 00000000000..d194fbaaafe
>> --- /dev/null
>> +++ b/sysdeps/unix/sysv/linux/tst-mseal-pkey.c
>> @@ -0,0 +1,84 @@
>> +/* Basic tests for mseal and pkey.
>> +   Copyright (C) 2025 Free Software Foundation, Inc.
>> +   This file is part of the GNU C Library.
>> +
>> +   The GNU C Library is free software; you can redistribute it and/or
>> +   modify it under the terms of the GNU Lesser General Public
>> +   License as published by the Free Software Foundation; either
>> +   version 2.1 of the License, or (at your option) any later version.
>> +
>> +   The GNU C Library is distributed in the hope that it will be useful,
>> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
>> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
>> +   Lesser General Public License for more details.
>> +
>> +   You should have received a copy of the GNU Lesser General Public
>> +   License along with the GNU C Library; if not, see
>> +   <https://www.gnu.org/licenses/>.  */
>> +
>> +#include <errno.h>
>> +#include <sys/mman.h>
>> +#include <support/check.h>
>> +#include <support/xunistd.h>
>> +
>> +static int
>> +do_test (void)
>> +{
>> +  TEST_VERIFY_EXIT (mseal (MAP_FAILED, 0, 0) == -1);
>> +  if (errno == ENOSYS || errno == EPERM)
>> +    FAIL_UNSUPPORTED ("kernel does not support mseal");
>> +  TEST_COMPARE (errno, EINVAL);
> 
> Shouldn't this be TEST_VERIFY instead of TEST_VERIFY_EXIT? Currently it
> would exit on 32-bit platforms and not check for mseal being
> unsupported.

Alright, I will change to TEST_VERIFY.

> 
>> +  long pagesize = xsysconf (_SC_PAGESIZE);
> 
> Any reason not to use getpagesize () like in the other test? Or should
> that one be changed to xsysconf (_SC_PAGESIZE)?
> 
>> +#include <support/test-driver.c>
>> diff --git a/sysdeps/unix/sysv/linux/tst-mseal.c b/sysdeps/unix/sysv/linux/tst-mseal.c
>> new file mode 100644
>> index 00000000000..0aff1e9e4c1
>> --- /dev/null
>> +++ b/sysdeps/unix/sysv/linux/tst-mseal.c
>> @@ -0,0 +1,67 @@
>> +/* Basic tests for mseal.
>> +   Copyright (C) 2025 Free Software Foundation, Inc.
>> +   This file is part of the GNU C Library.
>> +
>> +   The GNU C Library is free software; you can redistribute it and/or
>> +   modify it under the terms of the GNU Lesser General Public
>> +   License as published by the Free Software Foundation; either
>> +   version 2.1 of the License, or (at your option) any later version.
>> +
>> +   The GNU C Library is distributed in the hope that it will be useful,
>> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
>> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
>> +   Lesser General Public License for more details.
>> +
>> +   You should have received a copy of the GNU Lesser General Public
>> +   License along with the GNU C Library; if not, see
>> +   <https://www.gnu.org/licenses/>.  */
>> +
>> +#include <errno.h>
>> +#include <sys/mman.h>
>> +#include <support/check.h>
>> +#include <support/xunistd.h>
>> +
>> +static int
>> +do_test (void)
>> +{
>> +  TEST_VERIFY_EXIT (mseal (MAP_FAILED, 0, 0) == -1);
>> +  if (errno == ENOSYS || errno == EPERM)
>> +    FAIL_UNSUPPORTED ("kernel does not support mseal");
>> +  TEST_COMPARE (errno, EINVAL);
> 
> Likewise, I think this one should be TEST_VERIFY.

Ack.

> 
> Tested on x86_64 and things are fine, minus math failures seemingly due
> to a previous change.
> 
> This one should be good after minor local changes, so:
> 
> Reviewed-by: Collin Funk <collin.funk1@gmail.com>
> 
> Thanks,
> Collin