RFR [10] 8195823: Buffers given to response body subscribers should not contain unprocessed HTTP data

Daniel Fuchs daniel.fuchs at oracle.com
Mon Jan 22 18:40:36 UTC 2018


Hi Chris,

This looks good to me. I believed you covered all the
cases.

I have also imported your patch, and verified that the two new
tests you added were failing without the patch and passing
with the patch applied.

So if you ran that through mach5 I believe we're covered.

best regards,

-- daniel

On 22/01/2018 16:56, Chris Hegarty wrote:
> It has been observed that byte buffers passed to the response
> subscriber may contain unprocessed data in the area between
> the buffer's limit and its capacity. This can happen when:
>    1) reading chunked HTTP/1.1 response data
>    2) receiving HTTP/2 data frames for several streams
> 
> It is problematic for many reasons, least of all is that the
> asString and buffering subscribers accumulate bytes, below a
> certain threshold, into given buffers ( so as to not necessarily
> consumer memory ). There could also be other, more serious,
> issues.
> 
> http://cr.openjdk.java.net/~chegar/8195823/webrev.01/index.html
> 
> -Chris.
> 
> P.S. I will seek the required approval for 10 RDP 2 post code
> review.



More information about the net-dev mailing list