-32 bytes to copy, bounce through the stack */
> > > .Lrunt: movdqa %xmm1, bounce+16(%rsp) # clear out rest of on-
> > > stack copy
> > > - bts %r10d, %r8d # treat end of buffer as end of
> > > string
> > > - and %r9w, %r8w # end of string within first bu
> > > ffer?
> > > + bts %r10, %r8 # treat end of buffer as end of
> > > string
> > > + and %r9d, %r8d # mask out head before string
> > > + test $0x1ffff, %r8d # end of string within first ch
> > > unk or right after?
> > > jnz 0f # if yes, do not inspect second
> > > buffer
> > > 
> > > movdqa 16(%rsi), %xmm0 # load second chunk of input
> > >
> >
> > I've opened PR/291720 regarding a significant regression caused by this 
> > commit. It affects my older machines, resulting in enviornment (getenv) 
> > corruption. It does not affect my newer (and with more RAM) machines.
> 
> 
> BTW, this brought my postfix server effectively offline. I have not been 
> able to send or receive email until this commit was reverted locally.
> 
> 
> 
> 
> -- 
> Cheers,
> Cy Schubert <Cy.Schubert@cschubert.com <mailto:Cy.Schubert@cschubert.com>>
> FreeBSD UNIX: <cy@FreeBSD.org <mailto:cy@FreeBSD.org>> Web: https://FreeBSD.org <https://FreeBSD.org>
> NTP: <cy@nwtime.org <mailto:cy@nwtime.org>> Web: https://nwtime.org <https://nwtime.org>
> 
> 
> e**(i*pi)+1=0
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 

-- 
()  ascii ribbon campaign - for an encoding-agnostic world
/\  - against html email  - against proprietary attachments