Fix hstore hash function for empty hstores upgraded from 8.4.

Hstore data generated on pg 8.4 and pg_upgraded to current versions
remains in its original on-disk format unless modified. The same goes
for values generated by the addon hstore-new module on pre-9.0
versions. (The hstoreUpgrade function converts old values on the fly
when read in, but the on-disk value is not modified by this.)

Since old-format empty hstores (and hstore-new hstores) have
representations compatible with the new format, hstoreUpgrade thought
it could get away without modifying such values; but this breaks
hstore_hash (and the new hstore_hash_extended) which assumes
bit-perfect matching between semantically identical hstore values.
Only one bit actually differs (the "new version" flag in the count
field) but that of course is enough to break the hash.

Fix by making hstoreUpgrade unconditionally convert all old values to
new format.

Backpatch all the way, even though this changes a hash value in some
cases, because in those cases the hash value is already failing - for
example, a hash join between old- and new-format empty hstores will be
failing to match, or a hash index on an hstore column containing an
old-format empty value will be failing to find the value since it will
be searching for a hash derived from a new-format datum. (There are no
known field reports of this happening, probably because hashing of
hstores has only been useful in limited circumstances and there
probably isn't much upgraded data being used this way.)

Per concerns arising from discussion of commit eb6f29141b. Original
bug is my fault.

Discussion: https://postgr.es/m/60b1fd3b-7332-40f0-7e7f-f2f04f777747%402ndquadrant.com
This commit is contained in:
Andrew Gierth 2018-11-24 09:59:49 +00:00
parent 757c5182f2
commit d5890f49da
1 changed files with 20 additions and 29 deletions

View File

@ -238,34 +238,35 @@ hstoreUpgrade(Datum orig)
HStore *hs = (HStore *) PG_DETOAST_DATUM(orig);
int valid_new;
int valid_old;
bool writable;
/* Return immediately if no conversion needed */
if ((hs->size_ & HS_FLAG_NEWVERSION) ||
hs->size_ == 0 ||
(VARSIZE(hs) < 32768 && HSE_ISFIRST((ARRPTR(hs)[0]))))
if (hs->size_ & HS_FLAG_NEWVERSION)
return hs;
/* Do we have a writable copy? If not, make one. */
if ((void *) hs == (void *) DatumGetPointer(orig))
hs = (HStore *) PG_DETOAST_DATUM_COPY(orig);
if (hs->size_ == 0 ||
(VARSIZE(hs) < 32768 && HSE_ISFIRST((ARRPTR(hs)[0]))))
{
HS_SETCOUNT(hs, HS_COUNT(hs));
HS_FIXSIZE(hs, HS_COUNT(hs));
return hs;
}
valid_new = hstoreValidNewFormat(hs);
valid_old = hstoreValidOldFormat(hs);
/* Do we have a writable copy? */
writable = ((void *) hs != (void *) DatumGetPointer(orig));
if (!valid_old || hs->size_ == 0)
{
if (valid_new)
{
/*
* force the "new version" flag and the correct varlena length,
* but only if we have a writable copy already (which we almost
* always will, since short new-format values won't come through
* here)
* force the "new version" flag and the correct varlena length.
*/
if (writable)
{
HS_SETCOUNT(hs, HS_COUNT(hs));
HS_FIXSIZE(hs, HS_COUNT(hs));
}
HS_SETCOUNT(hs, HS_COUNT(hs));
HS_FIXSIZE(hs, HS_COUNT(hs));
return hs;
}
else
@ -302,15 +303,10 @@ hstoreUpgrade(Datum orig)
elog(WARNING, "ambiguous hstore value resolved as hstore-new");
/*
* force the "new version" flag and the correct varlena length, but
* only if we have a writable copy already (which we almost always
* will, since short new-format values won't come through here)
* force the "new version" flag and the correct varlena length.
*/
if (writable)
{
HS_SETCOUNT(hs, HS_COUNT(hs));
HS_FIXSIZE(hs, HS_COUNT(hs));
}
HS_SETCOUNT(hs, HS_COUNT(hs));
HS_FIXSIZE(hs, HS_COUNT(hs));
return hs;
#else
elog(WARNING, "ambiguous hstore value resolved as hstore-old");
@ -318,13 +314,8 @@ hstoreUpgrade(Datum orig)
}
/*
* must have an old-style value. Overwrite it in place as a new-style one,
* making sure we have a writable copy first.
* must have an old-style value. Overwrite it in place as a new-style one.
*/
if (!writable)
hs = (HStore *) PG_DETOAST_DATUM_COPY(orig);
{
int count = hs->size_;
HEntry *new_entries = ARRPTR(hs);