Postgres Pro
Downloads
Home   >   mailing lists

Thread: Potential issue in listExtensions()

Potential issue in listExtensions()

From
Suraj Kharage
Date:
Hi,

Upstream commit d696406a9b255546bc1716d07199103abd8bb785 [1] added the support for default extension version in \dx output and changed the query that fetches the extension list.
The changed query seems problematic and might display duplicate entries of extension if the same object oid exists in pg_description.

If I understand correctly, after oid wraparound, it is possible that the existing oid from another catalog might be used again for other catalog objects as per GetNewOidWithIndex().
If this is true, then it is possible that oid exists in pg_description for an object will be used for extension oid, and we might get a duplicate entry in \dx output.

```
"FROM pg_catalog.pg_extension e "
 "LEFT JOIN pg_catalog.pg_namespace n ON n.oid = e.extnamespace "
 "LEFT JOIN pg_catalog.pg_description d ON d.objoid = e.oid "
 "LEFT JOIN pg_catalog.pg_available_extensions() ae(name, default_version, comment) ON ae.name = e.extname "
 "AND d.classoid = 'pg_catalog.pg_extension'::pg_catalog.regclass\n",
```

We may need to handle this by moving the AND condition to the left join for pg_catalog.pg_description, so that we retrieve only pg_extension entries from pg_description.
Attaching the patch for the same.

Thoughts?

--

Thanks & Regards, 
Suraj kharage, 

Attachment

Re: Potential issue in listExtensions()

From
Dilip Kumar
Date:
On Thu, Jun 5, 2025 at 5:53 AM Suraj Kharage
<suraj.kharage@enterprisedb.com> wrote:
>
> Hi,
>
> Upstream commit d696406a9b255546bc1716d07199103abd8bb785 [1] added the support for default extension version in \dx
outputand changed the query that fetches the extension list. 
> The changed query seems problematic and might display duplicate entries of extension if the same object oid exists in
pg_description.
>
> If I understand correctly, after oid wraparound, it is possible that the existing oid from another catalog might be
usedagain for other catalog objects as per GetNewOidWithIndex(). 
> If this is true, then it is possible that oid exists in pg_description for an object will be used for extension oid,
andwe might get a duplicate entry in \dx output. 

Yes your understanding is correct and there is a possibility of the
bug you mentioned.

> ```
> "FROM pg_catalog.pg_extension e "
>  "LEFT JOIN pg_catalog.pg_namespace n ON n.oid = e.extnamespace "
>  "LEFT JOIN pg_catalog.pg_description d ON d.objoid = e.oid "
>  "LEFT JOIN pg_catalog.pg_available_extensions() ae(name, default_version, comment) ON ae.name = e.extname "
>  "AND d.classoid = 'pg_catalog.pg_extension'::pg_catalog.regclass\n",
> ```
>
> We may need to handle this by moving the AND condition to the left join for pg_catalog.pg_description, so that we
retrieveonly pg_extension entries from pg_description. 
> Attaching the patch for the same.
>
> Thoughts?
>
Attached patch seems to be fixing the issue.



--
Regards,
Dilip Kumar
Google

Re: Potential issue in listExtensions()

From
Magnus Hagander
Date:


On Thu, Jun 5, 2025 at 9:50 AM Dilip Kumar <dilipbalaut@gmail.com> wrote:
On Thu, Jun 5, 2025 at 5:53 AM Suraj Kharage
<suraj.kharage@enterprisedb.com> wrote:
>
> Hi,
>
> Upstream commit d696406a9b255546bc1716d07199103abd8bb785 [1] added the support for default extension version in \dx output and changed the query that fetches the extension list.
> The changed query seems problematic and might display duplicate entries of extension if the same object oid exists in pg_description.
>
> If I understand correctly, after oid wraparound, it is possible that the existing oid from another catalog might be used again for other catalog objects as per GetNewOidWithIndex().
> If this is true, then it is possible that oid exists in pg_description for an object will be used for extension oid, and we might get a duplicate entry in \dx output.

Yes your understanding is correct and there is a possibility of the
bug you mentioned.

Yup, agreed. Simply, the new LEFT JOIN was added on the wrong line. 

> ```
> "FROM pg_catalog.pg_extension e "
>  "LEFT JOIN pg_catalog.pg_namespace n ON n.oid = e.extnamespace "
>  "LEFT JOIN pg_catalog.pg_description d ON d.objoid = e.oid "
>  "LEFT JOIN pg_catalog.pg_available_extensions() ae(name, default_version, comment) ON ae.name = e.extname "
>  "AND d.classoid = 'pg_catalog.pg_extension'::pg_catalog.regclass\n",
> ```
>
> We may need to handle this by moving the AND condition to the left join for pg_catalog.pg_description, so that we retrieve only pg_extension entries from pg_description.
> Attaching the patch for the same.
>
> Thoughts?
>
Attached patch seems to be fixing the issue.

LGTM as well, applied. 

--

Re: Potential issue in listExtensions()

From
Suraj Kharage
Date:
Thanks, Magnus and Dilip.
--

Thanks & Regards, 
Suraj kharage, 



On Thu, Jun 5, 2025 at 1:34 PM Magnus Hagander <magnus@hagander.net> wrote:


On Thu, Jun 5, 2025 at 9:50 AM Dilip Kumar <dilipbalaut@gmail.com> wrote:
On Thu, Jun 5, 2025 at 5:53 AM Suraj Kharage
<suraj.kharage@enterprisedb.com> wrote:
>
> Hi,
>
> Upstream commit d696406a9b255546bc1716d07199103abd8bb785 [1] added the support for default extension version in \dx output and changed the query that fetches the extension list.
> The changed query seems problematic and might display duplicate entries of extension if the same object oid exists in pg_description.
>
> If I understand correctly, after oid wraparound, it is possible that the existing oid from another catalog might be used again for other catalog objects as per GetNewOidWithIndex().
> If this is true, then it is possible that oid exists in pg_description for an object will be used for extension oid, and we might get a duplicate entry in \dx output.

Yes your understanding is correct and there is a possibility of the
bug you mentioned.

Yup, agreed. Simply, the new LEFT JOIN was added on the wrong line. 

> ```
> "FROM pg_catalog.pg_extension e "
>  "LEFT JOIN pg_catalog.pg_namespace n ON n.oid = e.extnamespace "
>  "LEFT JOIN pg_catalog.pg_description d ON d.objoid = e.oid "
>  "LEFT JOIN pg_catalog.pg_available_extensions() ae(name, default_version, comment) ON ae.name = e.extname "
>  "AND d.classoid = 'pg_catalog.pg_extension'::pg_catalog.regclass\n",
> ```
>
> We may need to handle this by moving the AND condition to the left join for pg_catalog.pg_description, so that we retrieve only pg_extension entries from pg_description.
> Attaching the patch for the same.
>
> Thoughts?
>
Attached patch seems to be fixing the issue.

LGTM as well, applied. 

--