Re: [Qemu-devel] [PATCH for-1.2 v3 2/3] json-parser: don't replicate tokens at each level of recursion

[Michael Roth]

On Wed, Aug 15, 2012 at 02:04:52PM -0600, Eric Blake wrote:
> On 08/15/2012 12:45 PM, Michael Roth wrote:
> > Currently, when parsing a stream of tokens we make a copy of the token
> > list at the beginning of each level of recursion so that we do not
> > modify the original list in cases where we need to fall back to an
> > earlier state.
> > 
> > In the worst case, we will only read 1 or 2 tokens off the list before
> > recursing again, which means an upper bound of roughly N^2 token 
> > allocations.
> > 
> > For a "reasonably" sized QMP request (in this a QMP representation of
> > cirrus_vga's device state, generated via QIDL, being passed in via
> > qom-set), this caused my 16GB's of memory to be exhausted before any
> > noticeable progress was made by the parser.
> > 
> > This patch works around the issue by using single copy of the token list
> > in the form of an indexable array so that we can save/restore state by
> > manipulating indices.
> > 
> > A subsequent commit adds a "large_dict" test case which exhibits the
> > same behavior as above. With this patch applied the test case successfully
> > completes in under a second.
> > 
> > Tested with valgrind, make check, and QMP.
> > 
> > Signed-off-by: Michael Roth <address@hidden>
> > ---
> >  json-parser.c |  230 
> > +++++++++++++++++++++++++++++++++++----------------------
> >  1 file changed, 142 insertions(+), 88 deletions(-)
> 
> I'm not the most familiar with this code, so take my review with a grain
> of salt, but I read through it and the transformation looks sane (and my
> non-code findings from v2 were fixed).
> 
> Reviewed-by: Eric Blake <address@hidden>
> 
> > +static JSONParserContext parser_context_save(JSONParserContext *ctxt)
> > +{
> > +    JSONParserContext saved_ctxt = {0};
> > +    saved_ctxt.tokens.pos = ctxt->tokens.pos;
> > +    saved_ctxt.tokens.count = ctxt->tokens.count;
> > +    saved_ctxt.tokens.buf = ctxt->tokens.buf;
> 
> Is it any simpler to condense 3 lines to 1:
> 
> saved_cts.tokens = ctxt->tokens;
> 
> > +    return saved_ctxt;
> > +}
> > +
> > +static void parser_context_restore(JSONParserContext *ctxt,
> > +                                   JSONParserContext saved_ctxt)
> > +{
> > +    ctxt->tokens.pos = saved_ctxt.tokens.pos;
> > +    ctxt->tokens.count = saved_ctxt.tokens.count;
> > +    ctxt->tokens.buf = saved_ctxt.tokens.buf;
> 
> and again, ctxt->tokens = saved_ctxt.tokens;

Poor function naming: save/restore apply to the token state, the other
fields in ctxt are unused, so I opted to set the fields explicitly.

Can probably make this read a little better by breaking token state off
into it's own struct, but I think we can clean that up later.

Thanks for the review!

> 
> -- 
> Eric Blake   address@hidden    +1-919-301-3266
> Libvirt virtualization library http://libvirt.org
>