Hello,

the manpage for stat, fstat, lstat and fstatat specify that the error
EOVERLFLOW means:

      pathname or fd refers to a file whose  size,  inode  number,  or
      number  of  blocks  cannot  be represented in, respectively, the
      types off_t, ino_t, or blkcnt_t.  This error can occur when, for
      example,  an  application  compiled on a 32-bit platform without
      -D_FILE_OFFSET_BITS=64 calls stat() on a file whose size exceeds
      (1<<31)-1 bytes.

However at least in libc6-2.35 and later EOVERFLOW is also returned if
st_atim.tv_sec, st_mtim.tv_sec or st_ctim.tv_sec don't fit into an
int32_t on platforms with __TIME_SIZE == 32.

Best regards
Uwe

Hello,

I did some more research about this issue, so I'm adding some more
information.
fstatat with __fstatat64_time64") in glibc, that's between 2.33.9000 and
2.34. Commit aa03f722f3b994aaf81e72a8904bf33196780930 that was added for
2.33 is also slightly relevant (for platforms with LFS but 32 bit
time_t).

There are a few more calls that can return an undocumented EOVERFLOW:

 - clock_gettime
   (since commit ec138c67cbda ("sysdeps/clock_gettime: Use clock_gettime64 if avaliable")
   that is already in glibc 2.31)

 - ftime
   (since commit 5d8aa97da233 ("time: Add 64-bit time_t support for ftime")
   that is included in glibc 2.33)

 - gettimeofday
   (since commit 7455b700279e ("y2038: linux: Provide __gettimeofday64 implementation")
   that is included in glibc 2.31)

 - sched_rr_get_interval
   (since commit b112f53e9d0f ("y2038: linux: Provide __sched_rr_get_interval64 implementation")
   that is included in glibc 2.32)

 - time
   (since commit 75c4044b9a49 ("y2038: linux: Provide __time64 implementation")
   included in glibc 2.33)

 - timespec_get
   (since commit f1c314d27552 ("y2038: linux: Provide __timespec_get64 implementation")
   included in glibc 2.32)
   I didn't find documenation about this function in manpages-dev, seems
   to have appeard for isoc11.

Best regards
Uwe

Hello,

I did some more research about this issue, so I'm adding some more
information.
fstatat with __fstatat64_time64") in glibc, that's between 2.33.9000 and
2.34. Commit aa03f722f3b994aaf81e72a8904bf33196780930 that was added for
2.33 is also slightly relevant (for platforms with LFS but 32 bit
time_t).

There are a few more calls that can return an undocumented EOVERFLOW:

 - clock_gettime
   (since commit ec138c67cbda ("sysdeps/clock_gettime: Use clock_gettime64 if avaliable")
   that is already in glibc 2.31)

 - ftime
   (since commit 5d8aa97da233 ("time: Add 64-bit time_t support for ftime")
   that is included in glibc 2.33)

 - gettimeofday
   (since commit 7455b700279e ("y2038: linux: Provide __gettimeofday64 implementation")
   that is included in glibc 2.31)

 - sched_rr_get_interval
   (since commit b112f53e9d0f ("y2038: linux: Provide __sched_rr_get_interval64 implementation")
   that is included in glibc 2.32)

 - time
   (since commit 75c4044b9a49 ("y2038: linux: Provide __time64 implementation")
   included in glibc 2.33)

 - timespec_get
   (since commit f1c314d27552 ("y2038: linux: Provide __timespec_get64 implementation")
   included in glibc 2.32)
   I didn't find documenation about this function in manpages-dev, seems
   to have appeard for isoc11.

Best regards
Uwe