[PATCH] chg: handle EOF reading data block

[Yuya Nishihara]

On Mon, 18 Jul 2016 19:34:10 +0100, Jun Wu wrote:
> # HG changeset patch
> # User Jun Wu <quark at fb.com>
> # Date 1468864506 -3600
> #      Mon Jul 18 18:55:06 2016 +0100
> # Node ID c063823d664a978415bde11e0e13a6c3983d1dd9
> # Parent  953839de96ab574caa40557c542c262286c6287c
> # Available At https://bitbucket.org/quark-zju/hg-draft
> #              hg pull https://bitbucket.org/quark-zju/hg-draft -r c063823d664a
> chg: handle EOF reading data block

> --- a/contrib/chg/hgclient.c
> +++ b/contrib/chg/hgclient.c
> @@ -126,10 +126,15 @@ static void readchannel(hgclient_t *hgc)
>  		return;  /* assumes input request */
>  
>  	size_t cursize = 0;
> +	int emptycount = 0;
>  	while (cursize < hgc->ctx.datasize) {
>  		rsize = recv(hgc->sockfd, hgc->ctx.data + cursize,
>  			     hgc->ctx.datasize - cursize, 0);
> -		if (rsize < 0)
> +		/* rsize == 0 normally indicates EOF, while it's also a valid
> +		 * packet size for unix socket. treat it as EOF and abort if
> +		 * we get many empty responses in a row. */
> +		emptycount = (rsize == 0 ? emptycount + 1 : 0);
> +		if (rsize < 0 || emptycount > 20)

I'm curious when blocking recv() may return 0 other than EOF and zero-size
request. I suspect that the original condition "rsize < 0" was totally wrong.