Skip site navigation (1) Skip section navigation (2)

Re: 64-bit API for large object

From: Nozomi Anzai <anzai(at)sraoss(dot)co(dot)jp>
To: pgsql-hackers(at)postgresql(dot)org
Subject: Re: 64-bit API for large object
Date: 2012-10-01 07:28:16
Message-ID: 20121001162816.ff3f3186d35c339d2b4e05bb@sraoss.co.jp (view raw or flat)
Thread:
Lists: pgsql-hackers
Here is 64-bit API for large object version 3 patch.

> I checked this patch. It looks good, but here are still some points to be
> discussed.
> 
> * I have a question. What is the meaning of INT64_IS_BUSTED?
>   It seems to me a marker to indicate a platform without 64bit support.
>   However, the commit 901be0fad4034c9cf8a3588fd6cf2ece82e4b8ce
>   says as follows:
>   | Remove all the special-case code for INT64_IS_BUSTED, per decision that
>   | we're not going to support that anymore.

Removed INT64_IS_BUSTED.


> * At inv_seek(), it seems to me it checks offset correctness with wrong way,
>   as follows:
> |          case SEEK_SET:
> |              if (offset < 0)
> |                  elog(ERROR, "invalid seek offset: " INT64_FORMAT, offset);
> |              obj_desc->offset = offset;
> |              break;
>   It is a right assumption, if large object size would be restricted to 2GB.
>   But the largest positive int64 is larger than expected limitation.
>   So, it seems to me it should be compared with (INT_MAX * PAGE_SIZE)
>   instead.

Fixed.


> * At inv_write(), it definitely needs a check to prevent data-write upper 4TB.
>   In case when obj_desc->offset is a bit below 4TB, an additional 1GB write
>   will break head of the large object because of "pageno" overflow.

Added a such check.


> * Please also add checks on inv_read() to prevent LargeObjectDesc->offset
>   unexpectedly overflows 4TB boundary.

Added a such check.


> * At inv_truncate(), variable "off" is re-defined to int64. Is it really needed
>   change? All its usage is to store the result of "len % LOBLKSIZE".

Fixed and back to int32.


> Thanks,
> 
> 2012/9/24 Nozomi Anzai <anzai(at)sraoss(dot)co(dot)jp>:
> > Here is 64-bit API for large object version 2 patch.
> >
> >> I checked this patch. It can be applied onto the latest master branch
> >> without any problems. My comments are below.
> >>
> >> 2012/9/11 Tatsuo Ishii <ishii(at)postgresql(dot)org>:
> >> > Ok, here is the patch to implement 64-bit API for large object, to
> >> > allow to use up to 4TB large objects(or 16TB if BLCKSZ changed to
> >> > 32KB). The patch is based on Jeremy Drake's patch posted on September
> >> > 23, 2005
> >> > (http://archives.postgresql.org/pgsql-hackers/2005-09/msg01026.php)
> >> > and reasonably updated/edited to adopt PostgreSQL 9.3 by Nozomi Anzai
> >> > for the backend part and Yugo Nagata for the rest(including
> >> > documentation patch).
> >> >
> >> > Here are changes made in the patch:
> >> >
> >> > 1) Frontend lo_* libpq functions(fe-lobj.c)(Yugo Nagata)
> >> >
> >> > lo_initialize() gathers backend 64-bit large object handling
> >> > function's oid, namely lo_lseek64, lo_tell64, lo_truncate64.
> >> >
> >> > If client calls lo_*64 functions and backend does not support them,
> >> > lo_*64 functions return error to caller. There might be an argument
> >> > since calls to lo_*64 functions can automatically be redirected to
> >> > 32-bit older API. I don't know this is worth the trouble though.
> >> >
> >> I think it should definitely return an error code when user tries to
> >> use lo_*64 functions towards the backend v9.2 or older, because
> >> fallback to 32bit API can raise unexpected errors if application
> >> intends to seek the area over than 2GB.
> >>
> >> > Currently lo_initialize() throws an error if one of oids are not
> >> > available. I doubt we do the same way for 64-bit functions since this
> >> > will make 9.3 libpq unable to access large objects stored in pre-9.2
> >> > PostgreSQL servers.
> >> >
> >> It seems to me the situation to split the case of pre-9.2 and post-9.3
> >> using a condition of "conn->sversion >= 90300".
> >
> > Fixed so, and tested it by deleteing the lo_tell64's row from pg_proc.
> >
> >
> >> > To pass 64-bit integer to PQfn, PQArgBlock is used like this: int *ptr
> >> > is a pointer to 64-bit integer and actual data is placed somewhere
> >> > else. There might be other way: add new member to union u to store
> >> > 64-bit integer:
> >> >
> >> > typedef struct
> >> > {
> >> >         int                     len;
> >> >         int                     isint;
> >> >         union
> >> >         {
> >> >                 int                *ptr;                /* can't use void (dec compiler barfs)   */
> >> >                 int                     integer;
> >> >                 int64           bigint;         /* 64-bit integer */
> >> >         }                       u;
> >> > } PQArgBlock;
> >> >
> >> > I'm a little bit worried about this way because PQArgBlock is a public
> >> > interface.
> >> >
> >> I'm inclined to add a new field for the union; that seems to me straight
> >> forward approach.
> >> For example, the manner in lo_seek64() seems to me confusable.
> >> It set 1 on "isint" field even though pointer is delivered actually.
> >>
> >> +       argv[1].isint = 1;
> >> +       argv[1].len = 8;
> >> +       argv[1].u.ptr = (int *) &len;
> >
> > Your proposal was not adopted per discussion.
> >
> >
> >> > Also we add new type "pg_int64":
> >> >
> >> > #ifndef NO_PG_INT64
> >> > #define HAVE_PG_INT64 1
> >> > typedef long long int pg_int64;
> >> > #endif
> >> >
> >> > in postgres_ext.h per suggestion from Tom Lane:
> >> > http://archives.postgresql.org/pgsql-hackers/2005-09/msg01062.php
> >> >
> >> I'm uncertain about context of this discussion.
> >>
> >> Does it make matter if we include <stdint.h> and use int64_t instead
> >> of the self defined data type?
> >
> > Your proposal was not adopted per discussion.
> > Per discussion, endiannness translation was moved to fe-lobj.c.
> >
> >
> >> > 2) Backend lo_* functions (be-fsstubs.c)(Nozomi Anzai)
> >> >
> >> > Add lo_lseek64, lo_tell64, lo_truncate64 so that they can handle
> >> > 64-bit seek position and data length. loread64 and lowrite64 are not
> >> > added because if a program tries to read/write more than 2GB at once,
> >> > it would be a sign that the program need to be re-designed anyway.
> >> >
> >> I think it is a reasonable.
> >>
> >> > 3) Backend inv_api.c functions(Nozomi Anzai)
> >> >
> >> > No need to add new functions. Just extend them to handle 64-bit data.
> >> >
> >> > BTW , what will happen if older 32-bit libpq accesses large objects
> >> > over 2GB?
> >> >
> >> > lo_read and lo_write: they can read or write lobjs using 32-bit API as
> >> > long as requested read/write data length is smaller than 2GB. So I
> >> > think we can safely allow them to access over 2GB lobjs.
> >> >
> >> > lo_lseek: again as long as requested offset is smaller than 2GB, there
> >> > would be no problem.
> >> >
> >> > lo_tell:if current seek position is beyond 2GB, returns an error.
> >> >
> >> Even though iteration of lo_lseek() may move the offset to 4TB, it also
> >> makes unavailable to use lo_tell() to obtain the current offset, so I think
> >> it is reasonable behavior.
> >>
> >> However, error code is not an appropriate one.
> >>
> >> +       if (INT_MAX < offset)
> >> +       {
> >> +               ereport(ERROR,
> >> +                               (errcode(ERRCODE_UNDEFINED_OBJECT),
> >> +                                errmsg("invalid large-object
> >> descriptor: %d", fd)));
> >> +               PG_RETURN_INT32(-1);
> >> +       }
> >>
> >> According to the manpage of lseek(2)
> >>     EOVERFLOW
> >>         The resulting file offset cannot be represented in an off_t.
> >>
> >> Please add a new error code such as ERRCODE_BLOB_OFFSET_OVERFLOW.
> >
> > Changed the error code and error message. We added a new error code
> > "ERRCODE_UNDEFINED_OBJECT (22P07)".
> >
> >
> >> > 4) src/test/examples/testlo64.c added for 64-bit API example(Yugo Nagata)
> >> >
> >> > Comments and suggestions are welcome.
> >> >
> >> miscellaneous comments are below.
> >>
> >> Regression test is helpful. Even though no need to try to create 4TB large
> >> object, it is helpful to write some chunks around the design boundary.
> >> Could you add some test cases that writes some chunks around 4TB offset.
> >
> > Added 64-bit lobj test items into regression test and confirmed it worked
> > rightly.
> >
> >
> >> Thanks,
> >> --
> >> KaiGai Kohei <kaigai(at)kaigai(dot)gr(dot)jp>
> >>
> >>
> >> --
> >> Sent via pgsql-hackers mailing list (pgsql-hackers(at)postgresql(dot)org)
> >> To make changes to your subscription:
> >> http://www.postgresql.org/mailpref/pgsql-hackers
> >
> >
> > --
> > Nozomi Anzai
> > SRA OSS, Inc. Japan
> >
> >
> > --
> > Sent via pgsql-hackers mailing list (pgsql-hackers(at)postgresql(dot)org)
> > To make changes to your subscription:
> > http://www.postgresql.org/mailpref/pgsql-hackers
> >
> 
> 
> 
> -- 
> KaiGai Kohei <kaigai(at)kaigai(dot)gr(dot)jp>
> 
> 
> -- 
> Sent via pgsql-hackers mailing list (pgsql-hackers(at)postgresql(dot)org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-hackers


-- 
Nozomi Anzai
SRA OSS, Inc. Japan

Attachment: lobj64-v3.patch
Description: application/octet-stream (34.8 KB)

In response to

Responses

pgsql-hackers by date

Next:From: Heikki LinnakangasDate: 2012-10-01 07:29:31
Subject: Re: ToDo: allow to get a number of processed rows by COPY statement [Review of Patch]
Previous:From: Hannu KrosingDate: 2012-10-01 07:11:24
Subject: Re: is JSON really "a type" (Re: data to json enhancements)

Privacy Policy | About PostgreSQL
Copyright © 1996-2014 The PostgreSQL Global Development Group