Re: Invalid memory alloc request size for repeat()
Tom Lane <tgl@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Japin Li <japinli@hotmail.com>
Cc: "pgsql-hackers@lists.postgresql.org" <pgsql-hackers@lists.postgresql.org>
Date: 2022-05-25T14:50:52Z
Lists: pgsql-hackers
Japin Li <japinli@hotmail.com> writes:
> Today, I try to use repeat() to generate 1GB text, and it occurs invalid memory
> alloc request size [1]. It is a limit from palloc(), then I try to reduce it,
> it still complains out of memory which comes from enlargeStringInfo() [2]. The
> documentation about repect() [3] doesn't mentaion the limitation.
It would probably make sense for repeat() to check this explicitly:
if (unlikely(pg_mul_s32_overflow(count, slen, &tlen)) ||
- unlikely(pg_add_s32_overflow(tlen, VARHDRSZ, &tlen)))
+ unlikely(pg_add_s32_overflow(tlen, VARHDRSZ, &tlen)) ||
+ unlikely(!AllocSizeIsValid(tlen)))
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("requested length too large")));
The failure in enlargeStringInfo is probably coming from trying to
construct an output message to send back to the client. That's
going to be a lot harder to do anything nice about (and even if
the backend didn't fail, the client might).
regards, tom lane
Commits
-
Avoid ERRCODE_INTERNAL_ERROR in oracle_compat.c functions.
- 6217053f4e85 15.0 landed