policykit build fails -- polkitagenthelperprivate.c uses fdatasync()

[gnw3]

In polkitagenthelperprivate.c there are two calls to fdatasync(). This results in a build failure (unsatfisfied external).

According the linux man page

On  POSIX  systems  on  which fdatasync() is available,
_POSIX_SYNCHRONIZED_IO is defined in <unistd.h> to a 
value greater than 0.

Here, Leopard and Snow Leopard /usr/include/unistd.h as well as the Developer/SDK versions all have:

#define _POSIX_SYNCHRONIZED_IO (-1)

As a workaround, I just edited the file to replace fdatasync with fdata and the build succeeded. A Google search suggests that some OS X systems do have fdatasync, but that the headers don't provide a prototype.

[ryandesign (Ryan Carsten Schmidt)]

The build succeeded for me on Snow Leopard, but not Leopard or Tiger, so that's what I wrote in the ​bug report I submitted to the developers.

[ryandesign (Ryan Carsten Schmidt)]

Has duplicate #29282.

[chris@…]

Cc Me!

[chris@…]

I think gnwiii may have meant he replaced fdatasync with fsync, not fdata. When I replaced with fdata I got an error, but with fsync it worked. Also, this page says fdatasync and fsync should be the same:

​http://pubs.opengroup.org/onlinepubs/009695399/functions/fdatasync.html

[ebbinghaus@…]

Cc Me!

[fiee (Henning Hraban Ramm)]

ACK: When I replace "fdatasync" with "fsync" in /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_security_policykit/work/polkit-0.101/src/polkitagent/polkitagenthelperprivate.c it works for me on OSX 10.5.8 Intel.

[gnw3]

Replying to chris@…:

I think gnwiii may have meant he replaced fdatasync with fsync, not fdata. When I replaced with fdata I got an error, but with fsync it worked. Also, this page says fdatasync and fsync should be the same:

​http://pubs.opengroup.org/onlinepubs/009695399/functions/fdatasync.html

You are right -- I meant to type "fsync", not "fdata". The two functions may not be the same, as fdatasync only claims "synchronized I/O data integrity completion", so fsync may be slower (I would think policykit should err on the side of portability over performance):

3.375 Synchronized I/O Data Integrity Completion

For read, [...].

For write, when the operation has been completed or diagnosed if unsuccessful. The write is complete only when the data specified in the write request is successfully transferred and all file system information required to retrieve the data is successfully transferred.

File attributes that are not necessary for data retrieval (access time, modification time, status change time) need not be successfully transferred prior to returning to the calling process.

3.376 Synchronized I/O File Integrity Completion

Identical to a synchronized I/O data integrity completion with the addition that all file attributes relative to the I/O operation (including access time, modification time, status change time) are successfully transferred prior to returning to the calling process.

http://pubs.opengroup.org/onlinepubs/009604599/basedefs/xbd_chap03.html

[ballapete (Peter "Pete" Dyballa)]

Cc Me!

[ballapete (Peter "Pete" Dyballa)]

Replying to hraban@…:

ACK: When I replace "fdatasync" with "fsync" in /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_security_policykit/work/polkit-0.101/src/polkitagent/polkitagenthelperprivate.c it works for me on OSX 10.5.8 Intel.

Same here with 10.5.8 PPC.

[blair (Blair Zajac)]

If someone wants to create a patch that applies the changes to polkitagenthelperprivate.c and modifies Portfile to apply the patch only on OSes older than Snow Leopard, I can commit it to MacPorts.

[dershow]

Cc Me!

[kgalinsky@…]

Call reinplace on polkitagenthelperprivate

[kgalinsky@…]

Attached a diff for a portfile that fixes the issue. It uses reinplace rather than applying a patch for older OSes (reinplace was already being run on other files for OSes older than Snow Leopard).

[McDutchie (Martijn Dekker)]

Cc Me!

[kgalinsky@…]

Fix as per blair's suggestion in case reinplace isn't suitable - includes patch for src/polkitagent/polkitagenthelperprivate.c and for the Portfile where the patch will only get applied for older OSes by using patchfiles-append in the pre-patch phase.

[kgalinsky@…]

Cc Me!

[giby_the_kid@…]

I tried to change fdatasync by fsync in /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_security_policykit/work/polkit-0.101/src/polkitagent/polkitagenthelperprivate.c it hasn't work...

[gnw3]

Replying to giby_the_kid@…:

I tried to change fdatasync by fsync in /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_security_policykit/work/polkit-0.101/src/polkitagent/polkitagenthelperprivate.c it hasn't work...

Not much to go here, but now that there are a couple proposed patches, why not ​create a local portfile repository and test one of the proposed fixes?

[ryandesign (Ryan Carsten Schmidt)]

Thanks; r78414.