Re: [PATCH] Initialize struct stat st_[acm]tim.tv_nsec when present

[Egmont Koblinger]

Hi guys,

It is a regression (on my Fedora at least), as timestamps were preserved
> with the earlier mc versions.
>

As far as I understand the current story, this is probably due to some
other change since mc didn't change its relevant code recently.

Also, generally it's not feasible to make a release each time a regression
is fixed. Each release fixes twenty bugs and introduces two new, that's the
way it goes :-D

That being said, this timestamp issue plus the pending mcview growing
issues (which are also quite notable usability problems) are I believe a
good reason to schedule a .20 release for the not-so-distant future (let's
say, within a month or so).

cheers,
egmont
___
mc-devel mailing list
https://mail.gnome.org/mailman/listinfo/mc-devel

Re: [PATCH] Initialize struct stat st_[acm]tim.tv_nsec when present

[Nerijus Baliunas]

On Sat, 6 May 2017 22:08:47 +0200 (CEST) "Yury V. Zaytsev"  
wrote:

> Regarding the release, I'm not sure about it. As far as I understand, it's 
> not a regression, but a bug that has been present for a very long time, so 
> if this is true, then we are not really in a hurry.

It is a regression (on my Fedora at least), as timestamps were preserved
with the earlier mc versions.

Regards,
Nerijus
___
mc-devel mailing list
https://mail.gnome.org/mailman/listinfo/mc-devel

Re: [PATCH] Initialize struct stat st_[acm]tim.tv_nsec when present

[Yury V. Zaytsev]

On Sat, 6 May 2017, Nerijus Baliunas wrote:

It works, and it fixes my yesterday reported problem with zip/rar 
subdirs. Thanks! IMHO it would be nice to release 4.8.20 with this fix, 
it is important.


Thanks for the confirmation, I have put the branch on review.

Regarding the release, I'm not sure about it. As far as I understand, it's 
not a regression, but a bug that has been present for a very long time, so 
if this is true, then we are not really in a hurry. But if Andrew would 
say we do have to release the next version once this fix is in, I will try 
to find time to start this process...


--
Sincerely yours,
Yury V. Zaytsev
___
mc-devel mailing list
https://mail.gnome.org/mailman/listinfo/mc-devel

Re: [PATCH] Initialize struct stat st_[acm]tim.tv_nsec when present

[Nerijus Baliunas]

On Sat, 6 May 2017 21:06:39 +0200 (CEST) "Yury V. Zaytsev"  
wrote:

> I've created a ticket and branch for this patch:
> 
>  https://midnight-commander.org/ticket/3821#comment:1
> 
> I've also patched all other places where stat is filled manually that I 
> could find. The all builds & tests seem to run through.
> 
> Does this version work, or problems remain?

It works, and it fixes my yesterday reported problem with zip/rar subdirs.
Thanks!
IMHO it would be nice to release 4.8.20 with this fix, it is important.

Regards,
Nerijus
___
mc-devel mailing list
https://mail.gnome.org/mailman/listinfo/mc-devel

Re: [PATCH] Initialize struct stat st_[acm]tim.tv_nsec when present

[Yury V. Zaytsev]

I've created a ticket and branch for this patch:

https://midnight-commander.org/ticket/3821#comment:1

I've also patched all other places where stat is filled manually that I 
could find. The all builds & tests seem to run through.


Does this version work, or problems remain?

On Wed, 19 Apr 2017, Andrey Gursky wrote:


struct stat in libc for Linux kernel contains few fields more since 14+
years [1].

from bits/stat.h:
   struct timespec st_atim;/* Time of last access.  */
   struct timespec st_mtim;/* Time of last modification.  */
   struct timespec st_ctim;/* Time of last status change.  */
# define st_atime st_atim.tv_sec/* Backward compatibility.  */
# define st_mtime st_mtim.tv_sec
# define st_ctime st_ctim.tv_sec

The conventional fields became an alias.

POSIX.1-2008 made struct stat st_[acm]tim mandatory [2].

OS takes care to initialize struct stat properly [3]. By not using an
OS syscall or a libc wrapper to fill struct stat, we have to take care
of initializing all fields (or at least those being used later) explicitly.

[1] https://www.sourceware.org/ml/libc-alpha/2002-12/msg00011.html
[2] https://www.sourceware.org/ml/libc-alpha/2009-11/msg00102.html
[3] https://www.sourceware.org/ml/libc-alpha/2002-12/msg00013.html

Fixes: file timestamps not preserved 
(https://mail.gnome.org/archives/mc-devel/2017-April/msg0.html)
Reported-By: Nerijus Baliunas 
---
configure.ac|  2 +-
src/vfs/cpio/cpio.c | 15 +++
src/vfs/tar/tar.c   |  3 +++
3 files changed, 19 insertions(+), 1 deletion(-)

diff --git a/configure.ac b/configure.ac
index 45836fcaf..387110daf 100644
--- a/configure.ac
+++ b/configure.ac
@@ -181,7 +181,7 @@ AC_TYPE_PID_T
AC_TYPE_UID_T

AC_STRUCT_ST_BLOCKS
-AC_CHECK_MEMBERS([struct stat.st_blksize, struct stat.st_rdev])
+AC_CHECK_MEMBERS([struct stat.st_blksize, struct stat.st_rdev, struct 
stat.st_mtim])
gl_STAT_SIZE

AH_TEMPLATE([sig_atomic_t],
diff --git a/src/vfs/cpio/cpio.c b/src/vfs/cpio/cpio.c
index dba1a07c3..ffe6267cf 100644
--- a/src/vfs/cpio/cpio.c
+++ b/src/vfs/cpio/cpio.c
@@ -462,9 +462,15 @@ cpio_create_entry (struct vfs_class *me, struct 
vfs_s_super *super, struct stat
entry->ino->st.st_mode = st->st_mode;
entry->ino->st.st_uid = st->st_uid;
entry->ino->st.st_gid = st->st_gid;
+#ifdef HAVE_STRUCT_STAT_ST_MTIM
+entry->ino->st.st_atim = st->st_atim;
+entry->ino->st.st_mtim = st->st_mtim;
+entry->ino->st.st_ctim = st->st_ctim;
+#else
entry->ino->st.st_atime = st->st_atime;
entry->ino->st.st_mtime = st->st_mtime;
entry->ino->st.st_ctime = st->st_ctime;
+#endif
}

g_free (name);
@@ -589,6 +595,9 @@ cpio_read_bin_head (struct vfs_class *me, struct 
vfs_s_super *super)
st.st_rdev = u.buf.c_rdev;
#endif
st.st_size = (u.buf.c_filesizes[0] << 16) | u.buf.c_filesizes[1];
+#ifdef HAVE_STRUCT_STAT_ST_MTIM
+st.st_atim.tv_nsec = st.st_mtim.tv_nsec = st.st_ctim.tv_nsec = 0;
+#endif
st.st_atime = st.st_mtime = st.st_ctime = (u.buf.c_mtimes[0] << 16) | 
u.buf.c_mtimes[1];

return cpio_create_entry (me, super, , name);
@@ -658,6 +667,9 @@ cpio_read_oldc_head (struct vfs_class *me, struct 
vfs_s_super *super)
u.st.st_rdev = hd.c_rdev;
#endif
u.st.st_size = hd.c_filesize;
+#ifdef HAVE_STRUCT_STAT_ST_MTIM
+u.st.st_atim.tv_nsec = u.st.st_mtim.tv_nsec = u.st.st_ctim.tv_nsec = 0;
+#endif
u.st.st_atime = u.st.st_mtime = u.st.st_ctime = hd.c_mtime;

return cpio_create_entry (me, super, , name);
@@ -736,6 +748,9 @@ cpio_read_crc_head (struct vfs_class *me, struct 
vfs_s_super *super)
u.st.st_rdev = makedev (hd.c_rdev, hd.c_rdevmin);
#endif
u.st.st_size = hd.c_filesize;
+#ifdef HAVE_STRUCT_STAT_ST_MTIM
+u.st.st_atim.tv_nsec = u.st.st_mtim.tv_nsec = u.st.st_ctim.tv_nsec = 0;
+#endif
u.st.st_atime = u.st.st_mtime = u.st.st_ctime = hd.c_mtime;

return cpio_create_entry (me, super, , name);
diff --git a/src/vfs/tar/tar.c b/src/vfs/tar/tar.c
index 3afd8d68d..4b674739d 100644
--- a/src/vfs/tar/tar.c
+++ b/src/vfs/tar/tar.c
@@ -442,6 +442,9 @@ tar_fill_stat (struct vfs_s_super *archive, struct stat 
*st, union record *heade
}

st->st_size = h_size;
+#ifdef HAVE_STRUCT_STAT_ST_MTIM
+st->st_atim.tv_nsec = st->st_mtim.tv_nsec = st->st_ctim.tv_nsec = 0;
+#endif
st->st_mtime = tar_from_oct (1 + 12, header->header.mtime);
st->st_atime = 0;
st->st_ctime = 0;
--
2.12.2+git20170411
___
mc-devel mailing list
https://mail.gnome.org/mailman/listinfo/mc-devel



--
Sincerely yours,
Yury V. Zaytsev
___
mc-devel mailing list
https://mail.gnome.org/mailman/listinfo/mc-devel

Re: file timestamps not preserved

[Yury V. Zaytsev]

On Sat, 6 May 2017, Nerijus Baliunas wrote:


On Fri, 5 May 2017 23:31:50 +0200 Andrey Gursky  wrote:


I've downloaded the original mc 4.8.19 release, applied the patches,
run autoreconf, then configure, make install and it works for me. Have
you tested with the file you've sent me? If yes, I'm curious what could
be else different?..

BTW, if I call autoconf instead of autoreconf, I cannot see anything
happens.


With autoreconf instead of autoconf it is OK. Thanks.


You have to use ./autogen.sh to bootstrap the build system.

--
Sincerely yours,
Yury V. Zaytsev
___
mc-devel mailing list
https://mail.gnome.org/mailman/listinfo/mc-devel

Re: [PATCH] Remove unused check in configure.ac

[Yury V. Zaytsev]

On Wed, 19 Apr 2017, Andrey Gursky wrote:


AC_STRUCT_ST_BLOCKS is a leftover from cleanup in 
0ba019a90b3798abae32ba261e72b737dc945615


I don't understand why you're saying that this macro is unused. It defines

HAVE_STRUCT_STAT_ST_BLOCKS

if struct `stat` contains an `st_blocks` member. This define is currently 
used in branches all over our codebase.


https://www.gnu.org/savannah-checkouts/gnu/autoconf/manual/autoconf-2.69/html_node/Particular-Structures.html#Particular-Structures


---
configure.ac | 1 -
1 file changed, 1 deletion(-)

diff --git a/configure.ac b/configure.ac
index 387110daf..ac81404f5 100644
--- a/configure.ac
+++ b/configure.ac
@@ -180,7 +180,6 @@ gl_PROMOTED_TYPE_MODE_T
AC_TYPE_PID_T
AC_TYPE_UID_T

-AC_STRUCT_ST_BLOCKS
AC_CHECK_MEMBERS([struct stat.st_blksize, struct stat.st_rdev, struct 
stat.st_mtim])
gl_STAT_SIZE

--
2.12.2+git20170411
___
mc-devel mailing list
https://mail.gnome.org/mailman/listinfo/mc-devel



--
Sincerely yours,
Yury V. Zaytsev
___
mc-devel mailing list
https://mail.gnome.org/mailman/listinfo/mc-devel

Re: file timestamps not preserved

[Nerijus Baliunas]

On Fri, 5 May 2017 23:31:50 +0200 Andrey Gursky  wrote:

> I've downloaded the original mc 4.8.19 release, applied the patches,
> run autoreconf, then configure, make install and it works for me. Have
> you tested with the file you've sent me? If yes, I'm curious what could
> be else different?..
> 
> BTW, if I call autoconf instead of autoreconf, I cannot see anything
> happens.

With autoreconf instead of autoconf it is OK. Thanks.

Regards,
Nerijus
___
mc-devel mailing list
https://mail.gnome.org/mailman/listinfo/mc-devel