git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH v2 3/3] sideband: add defense against packets missing a band designator

From
Johannes Schindelin <johannes.schindelin@gmx.de>
Date
Oct 23, 2020, 05:36 UTC
Message-ID
<nycvar.QRO.7.76.6.2010230732250.56@tvgsbejvaqbjf.bet>
In-Reply-To
<20201023084827.GE4012156@coredump.intra.peff.net>
Hi Peff,
On Fri, 23 Oct 2020, Jeff King wrote:
Show 21 quoted lines
> On Mon, Oct 19, 2020 at 07:35:42PM +0000, Johannes Schindelin via GitGitGadget wrote:
>
> > diff --git a/pkt-line.c b/pkt-line.c
> > index 657a702927..f72048f623 100644
> > --- a/pkt-line.c
> > +++ b/pkt-line.c
> > @@ -461,8 +461,10 @@ int recv_sideband(const char *me, int in_stream, int out)
> >  	enum sideband_type sideband_type;
> >
> >  	while (1) {
> > -		len = packet_read(in_stream, NULL, NULL, buf, LARGE_PACKET_MAX,
> > -				  0);
> > +		int status = packet_read_with_status(in_stream, NULL, NULL, buf,
> > +						     LARGE_PACKET_MAX, &len, 0);
> > +		if (!len && status == PACKET_READ_NORMAL)
> > +			BUG("missing band designator");
> >  		if (!demultiplex_sideband(me, buf, len, 0, &scratch,
> >  					  &sideband_type))
>
> I also wonder if this status-check should be pushed down into
> demultiplex_sideband() by passing "status",

I tried that, but as mentioned in the commit message, fbd76cd450 (sideband: reverse its dependency on pkt-line, 2019-01-16) went out of its way to _stop_ the code inside `demultiplex_sideband()` from relying on anything in `pkt-line.h`. And that `PACKET_READ_NORMAL` and `PACKET_READ_EOF` _is_ from `pkt-line.h`.

Ciao, Dscho

Show 61 quoted lines
> for two reasons:
>
>   1. So we don't have to repeat it (though it isn't that big)
>
>   2. The other half of this weirdness is that if we get an early EOF,
>      we'll hit the "missing sideband designator" die() message. But
>      that's not really what happened; we probably got a network hangup.
>      And we could distinguish that case by checking for status ==
>      PACKET_READ_EOF and provide a better message.
>
> Something like this (completely untested):
>
> diff --git a/sideband.c b/sideband.c
> index 0a60662fa6..6ad15ed581 100644
> --- a/sideband.c
> +++ b/sideband.c
> @@ -115,6 +115,7 @@ static void maybe_colorize_sideband(struct strbuf *dest, const char *src, int n)
>  #define DUMB_SUFFIX "        "
>
>  int demultiplex_sideband(const char *me, char *buf, int len,
> +			 enum packet_read_status status,
>  			 int die_on_error,
>  			 struct strbuf *scratch,
>  			 enum sideband_type *sideband_type)
> @@ -130,17 +131,29 @@ int demultiplex_sideband(const char *me, char *buf, int len,
>  			suffix = DUMB_SUFFIX;
>  	}
>
> -	if (len == 0) {
> -		*sideband_type = SIDEBAND_FLUSH;
> -		goto cleanup;
> -	}
> -	if (len < 1) {
> +	if (status == PACKET_READ_EOF) {
>  		strbuf_addf(scratch,
> -			    "%s%s: protocol error: no band designator",
> +			    "%s%s: protocol error: eof while reading packet",
>  			    scratch->len ? "\n" : "", me);
>  		*sideband_type = SIDEBAND_PROTOCOL_ERROR;
>  		goto cleanup;
>  	}
> +
> +	if (len < 0)
> +		BUG("negative length on non-eof packet read");
> +
> +	if (len == 0) {
> +		if (status == PACKET_READ_NORMAL) {
> +			strbuf_addf(scratch,
> +				    "%s%s protocol error: no band designator",
> +				    scratch->len ? "\n" : "", me);
> +			*sideband_type = SIDEBAND_PROTOCOL_ERROR;
> +		} else {
> +			*sideband_type = SIDEBAND_FLUSH;
> +		}
> +		goto cleanup;
> +	}
> +
>  	band = buf[0] & 0xff;
>  	buf[len] = '\0';
>  	len--;
>
Previous: Jeff KingNext: Jeff King
Message 19 of 33 in “t5500.43: make the check a bit more robust”
  1. t5500.43: make the check a bit more robustJohannes Schindelin via GitGitGadget, Oct 13, 2020
  2. Junio C HamanoOct 13, 2020
  3. Jeff KingOct 13, 2020
  4. Junio C HamanoOct 13, 2020
  5. Jeff KingOct 13, 2020
  6. Johannes SchindelinOct 17, 2020
  7. Johannes SchindelinOct 17, 2020
  8. 0/3 Work around flakiness in t5500.43Johannes Schindelin via GitGitGadget, Oct 19, 2020
  9. 1/3 sideband: avoid reporting incomplete sideband messagesJohannes Schindelin via GitGitGadget, Oct 19, 2020
  10. Junio C HamanoOct 20, 2020
  11. Johannes SchindelinOct 20, 2020
  12. Junio C HamanoOct 20, 2020
  13. 2/3 sideband: report unhandled incomplete sideband messages as bugsJohannes Schindelin via GitGitGadget, Oct 19, 2020
  14. 3/3 sideband: add defense against packets missing a band designatorJohannes Schindelin via GitGitGadget, Oct 19, 2020
  15. Junio C HamanoOct 20, 2020
  16. Jeff KingOct 23, 2020
  17. Junio C HamanoOct 23, 2020
  18. Jeff KingOct 23, 2020
  19. Johannes SchindelinOct 23, 2020
  20. Jeff KingOct 23, 2020
  21. Johannes SchindelinOct 26, 2020
  22. 0/2 Work around flakiness in t5500.43Johannes Schindelin via GitGitGadget, Oct 26, 2020
  23. 1/2 sideband: avoid reporting incomplete sideband messagesJohannes Schindelin via GitGitGadget, Oct 26, 2020
  24. 2/2 sideband: report unhandled incomplete sideband messages as bugsJohannes Schindelin via GitGitGadget, Oct 26, 2020
  25. Junio C HamanoOct 26, 2020
  26. sideband: diagnose more incoming packet anomaliesJeff King, Oct 27, 2020
  27. Jeff KingOct 27, 2020
  28. Junio C HamanoOct 27, 2020
  29. Jeff KingOct 27, 2020
  30. Junio C HamanoOct 27, 2020
  31. Jeff KingOct 28, 2020
  32. Jeff KingOct 27, 2020
  33. Junio C HamanoOct 27, 2020

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.