Re: longfin and tamandua aren't too happy but I'm not sure why - Mailing list pgsql-hackers

From Tom Lane
Subject Re: longfin and tamandua aren't too happy but I'm not sure why
Date
Msg-id 3837430.1664311847@sss.pgh.pa.us
Whole thread Raw
In response to Re: longfin and tamandua aren't too happy but I'm not sure why  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: longfin and tamandua aren't too happy but I'm not sure why
Re: longfin and tamandua aren't too happy but I'm not sure why
List pgsql-hackers
I wrote:
>   * frame #0: 0x000000010a36af8c postgres`ParseCommitRecord(info='\x80', xlrec=0x00007fa0678a8090,
parsed=0x00007ff7b5c50e78)at xactdesc.c:102:30 

Okay, so the problem is this: by widening RelFileNumber to 64 bits,
you have increased the alignment requirement of struct RelFileLocator,
and thereby also SharedInvalidationMessage, to 8 bytes where it had
been 4.  longfin's alignment check is therefore expecting that
xl_xact_twophase will likewise be 8-byte-aligned, but it isn't:

(lldb) p data
(char *) $0 = 0x00007fa06783a0a4 "\U00000001"

I'm not sure whether the code that generates commit WAL records is
breaking a contract it should maintain, or xactdesc.c needs to be
taught to not assume that this data is adequately aligned.

There is a second problem that I am going to hold your feet to the
fire about:

(lldb) p sizeof(SharedInvalidationMessage)
(unsigned long) $1 = 24

We have sweated a good deal for a long time to keep that struct
to 16 bytes.  I do not think 50% bloat is acceptable.

            regards, tom lane



pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: longfin and tamandua aren't too happy but I'm not sure why
Next
From: Robert Haas
Date:
Subject: Re: longfin and tamandua aren't too happy but I'm not sure why