Re: [dpdk-dev] [PATCH 2/3] examples/fips_validation: ignore \r in input files

[Olivier Matz <olivier.matz@6wind.com>]

Hi Fan,

So if we cannot know which version removed the \r, I suggest to just
drop this patch. I thought it was a bug in the parser, but if it does
not happen with files matching the supported CAVS version, there is
nothing to fix.

What do you think?

Thanks,
Olivier


On Thu, Oct 08, 2020 at 12:41:11PM +0000, Zhang, Roy Fan wrote:
> Hi Olivier,
> 
> Unfortunately I wanted to find the same document since forever. NIST
> did not provide this on their website. What I am sure is for CAVS 21.0
> both the test vectors Intel used for testing and the ones provided by
> our customer for debugging did not have \r in the files. In 2018 we
> could find some sample request and response files from NIST website
> but I just checked and they are gone.
> 
> Regards,
> Fan
> 
> > -----Original Message-----
> > From: Olivier Matz <olivier.matz@6wind.com>
> > Sent: Thursday, October 8, 2020 12:32 PM
> > To: Zhang, Roy Fan <roy.fan.zhang@intel.com>
> > Cc: dev@dpdk.org; Kovacevic, Marko <marko.kovacevic@intel.com>; Akhil
> > Goyal <akhil.goyal@nxp.com>; Kusztal, ArkadiuszX
> > <arkadiuszx.kusztal@intel.com>; stable@dpdk.org; Anoob Joseph
> > <anoobj@marvell.com>
> > Subject: Re: [PATCH 2/3] examples/fips_validation: ignore \r in input files
> > 
> > Hi Fan,
> > 
> > Thank you for the clarification. One more question: do you know where I
> > can find a description of the different FIPS CAVS versions? I would like
> > to know from what version the \r has been removed.
> > 
> > Thanks,
> > Olivier
> > 
> > On Thu, Oct 08, 2020 at 10:24:48AM +0000, Zhang, Roy Fan wrote:
> > > Hi Olivier,
> > >
> > > Sorry I didn't state myself clear in the first place.
> > >
> > > My intention is '\r' check, or any future CAVS version specific change to the
> > > application should be wrapped into a branch that is checked with parsed
> > > version number. With this way the original application's behavior should
> > > remain the same.
> > >
> > > The reason for that is we are having an issue right now that the validation
> > > team is struggling with the limited test vectors and inconsistency formatting
> > > between different FIPS CAVS versions. For example we still have FIPS TDES
> > test
> > > failing today due to the different test file versions.
> > > https://bugs.dpdk.org/show_bug.cgi?id=512
> > >
> > > The solution is certainly far from pretty but should help to share the
> > > maintenance effort amongst the contributors.
> > >
> > > The "FIPS_DEF_VERSION" can be removed of course.
> > >
> > > Regards,
> > > Fan
> > >
> > > > -----Original Message-----
> > > > From: Olivier Matz <olivier.matz@6wind.com>
> > > > Sent: Thursday, October 8, 2020 10:22 AM
> > > > To: Zhang, Roy Fan <roy.fan.zhang@intel.com>
> > > > Cc: dev@dpdk.org; Kovacevic, Marko <marko.kovacevic@intel.com>;
> > Akhil
> > > > Goyal <akhil.goyal@nxp.com>; Kusztal, ArkadiuszX
> > > > <arkadiuszx.kusztal@intel.com>; stable@dpdk.org; Anoob Joseph
> > > > <anoobj@marvell.com>
> > > > Subject: Re: [PATCH 2/3] examples/fips_validation: ignore \r in input files
> > > >
> > > > Hi,
> > > >
> > > > On Thu, Oct 08, 2020 at 08:50:25AM +0000, Zhang, Roy Fan wrote:
> > > > > Hi Olivier,
> > > > >
> > > > > Anood and us had the similar discussion.
> > > > >
> > > > > Can we change the sample application to parse version data instead,
> > > > > and for the version specific code changes we will wrap them by a
> > > > > branch to compare the parsed version and the expected version?
> > > > > (we probably should have done that long time ago).
> > > > >
> > > > > I drafted a code change to parse the version data, see if you think it
> > > > > is ok?
> > > >
> > > > Thank you for your feedback.
> > > >
> > > > The code that gets the version looks good to me (I just have a
> > > > small comment, see below). However I'm not sure what to do with it.
> > > >
> > > > Do you mean we should return an error if the version is incorrect? Or
> > > > should we only skip '\r' for old versions? FIPS_DEF_VERSION is not used
> > > > in your patch. In that case, I think it is a bit overkill. Do you think
> > > > it is a problem to always drop '\r'?
> > > >
> > > > If you think we should not support files containing '\r', I'm fine
> > > > with it, I can drop this particular patch.
> > > >
> > > >
> > > > >
> > > > > diff --git a/examples/fips_validation/fips_validation.c
> > > > b/examples/fips_validation/fips_validation.c
> > > > > index 9bdf257b8..9b6518c92 100644
> > > > > --- a/examples/fips_validation/fips_validation.c
> > > > > +++ b/examples/fips_validation/fips_validation.c
> > > > > @@ -98,7 +98,7 @@ fips_test_parse_header(void)
> > > > >  	uint32_t i;
> > > > >  	char *tmp;
> > > > >  	int ret;
> > > > > -	int algo_parsed = 0;
> > > > > +	int algo_parsed = 0, version_parsed = 0;
> > > > >  	time_t t = time(NULL);
> > > > >  	struct tm *tm_now = localtime(&t);
> > > > >
> > > > > @@ -107,6 +107,27 @@ fips_test_parse_header(void)
> > > > >  		return ret;
> > > > >
> > > > >  	for (i = 0; i < info.nb_vec_lines; i++) {
> > > > > +		/* parse the version info */
> > > > > +		tmp = strstr(info.vec[i], "CAVS ");
> > > > > +		if (tmp != NULL) {
> > > > > +			if (version_parsed != 0) {
> > > > > +				RTE_LOG(ERR, USER1,
> > > > > +					"Multiple version data\n");
> > > > > +				return -1;
> > > > > +			}
> > > > > +
> > > > > +			tmp = tmp + sizeof("CAVS ");
> > > >
> > > > I think it should be strlen(), because sizeof() will contain
> > > > the '\0'. Or it could be sizeof() - 1.
> > > >
> > > > > +
> > > > > +			if (strlen(tmp) >= MAX_VER_STRING_SIZE) {
> > > > > +				RTE_LOG(ERR, USER1, "Version (%s) too
> > > > long\n",
> > > > > +						tmp);
> > > > > +				return -1;
> > > > > +			}
> > > > > +
> > > > > +			strlcpy(info.version, tmp, MAX_VER_STRING_SIZE);
> > > > > +			version_parsed = 1;
> > > > > +		}
> > > > > +
> > > > >  		if (!algo_parsed) {
> > > > >  			if (strstr(info.vec[i], "AESVS")) {
> > > > >  				algo_parsed = 1;
> > > > > diff --git a/examples/fips_validation/fips_validation.h
> > > > b/examples/fips_validation/fips_validation.h
> > > > > index 75fa555fa..b8c60c55f 100644
> > > > > --- a/examples/fips_validation/fips_validation.h
> > > > > +++ b/examples/fips_validation/fips_validation.h
> > > > > @@ -15,6 +15,9 @@
> > > > >  #define MAX_BUF_SIZE		2048
> > > > >  #define MAX_STRING_SIZE		64
> > > > >  #define MAX_DIGEST_SIZE		64
> > > > > +#define MAX_VER_STRING_SIZE	8
> > > > > +
> > > > > +#define FIPS_DEF_VERSION	"21.0"
> > > > >
> > > > >  #define POSITIVE_TEST		0
> > > > >  #define NEGATIVE_TEST		-1
> > > > > @@ -155,6 +158,7 @@ struct sha_interim_data {
> > > > >  };
> > > > >
> > > > >  struct fips_test_interim_info {
> > > > > +	char version[MAX_VER_STRING_SIZE];
> > > > >  	FILE *fp_rd;
> > > > >  	FILE *fp_wr;
> > > > >  	enum file_types file_type;
> > > > >
> > > > >
> > > > > Regards,
> > > > > Fan
> > > > >
> > > > > > -----Original Message-----
> > > > > > From: Olivier Matz <olivier.matz@6wind.com>
> > > > > > Sent: Tuesday, October 6, 2020 11:09 AM
> > > > > > To: Zhang, Roy Fan <roy.fan.zhang@intel.com>
> > > > > > Cc: dev@dpdk.org; Kovacevic, Marko <marko.kovacevic@intel.com>;
> > > > Akhil
> > > > > > Goyal <akhil.goyal@nxp.com>; Kusztal, ArkadiuszX
> > > > > > <arkadiuszx.kusztal@intel.com>; stable@dpdk.org; Anoob Joseph
> > > > > > <anoobj@marvell.com>
> > > > > > Subject: Re: [PATCH 2/3] examples/fips_validation: ignore \r in input
> > files
> > > > > >
> > > > > > Hi Fan,
> > > > > >
> > > > > > On Tue, Oct 06, 2020 at 08:47:10AM +0000, Zhang, Roy Fan wrote:
> > > > > > > Hi Olivier,
> > > > > > >
> > > > > > > > -----Original Message-----
> > > > > > > > From: Olivier Matz <olivier.matz@6wind.com>
> > > > > > > > Sent: Tuesday, October 6, 2020 8:42 AM
> > > > > > > > To: dev@dpdk.org
> > > > > > > > Cc: Kovacevic, Marko <marko.kovacevic@intel.com>; Akhil Goyal
> > > > > > > > <akhil.goyal@nxp.com>; Zhang, Roy Fan
> > <roy.fan.zhang@intel.com>;
> > > > > > Kusztal,
> > > > > > > > ArkadiuszX <arkadiuszx.kusztal@intel.com>; stable@dpdk.org
> > > > > > > > Subject: [PATCH 2/3] examples/fips_validation: ignore \r in input
> > files
> > > > > > > >
> > > > > > > > Some test vectors contain '\r' before '\n' (see link). Ignore them.
> > > > > > > >
> > > > > > > > Link: https://www.openssl.org/docs/fips/testvectors-linux-2007-
> > 10-
> > > > > > 10.tar.gz
> > > > > > > > Fixes: 3d0fad56b74a ("examples/fips_validation: add crypto FIPS
> > > > > > application")
> > > > > > > > Cc: stable@dpdk.org
> > > > > > > >
> > > > > > > > Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
> > > > > > > > ---
> > > > > > > >  examples/fips_validation/fips_validation.c | 2 ++
> > > > > > > >  1 file changed, 2 insertions(+)
> > > > > > > >
> > > > > > > > diff --git a/examples/fips_validation/fips_validation.c
> > > > > > > > b/examples/fips_validation/fips_validation.c
> > > > > > > > index 13f763c9aa..858f581ba3 100644
> > > > > > > > --- a/examples/fips_validation/fips_validation.c
> > > > > > > > +++ b/examples/fips_validation/fips_validation.c
> > > > > > > > @@ -33,6 +33,8 @@ get_file_line(void)
> > > > > > > >
> > > > > > > >  		if (loc >= MAX_LINE_CHAR - 1)
> > > > > > > >  			return -ENOMEM;
> > > > > > > > +		if (c == '\r')
> > > > > > > > +			continue;
> > > > > > > >  		if (c == '\n')
> > > > > > > >  			break;
> > > > > > > >  		line[loc++] = c;
> > > > > > > > --
> > > > > > >
> > > > > > >
> > > > > > > The patch looks ok but the test file link you provided in the patch is
> > > > CAVS
> > > > > > > 5.3.
> > > > > > >
> > > > > > > As mentioned in
> > > > > > > https://doc.dpdk.org/guides/sample_app_ug/fips_validation.html,
> > the
> > > > > > supported
> > > > > > > CAVS supported version is 21.0 (not latest one by newer than 5.3).
> > In
> > > > CAVS
> > > > > > > 21.0 test files there is no '\r' before '\n' (I suppose this is for
> > Windows
> > > > > > > right).
> > > > > >
> > > > > > Thank you for your feedback.
> > > > > >
> > > > > > I'm ok to drop this patch from the patchset if you feel it's useless, or
> > > > > > I can update the commit log with the information you provide, to
> > clarify
> > > > > > that it should not happen with the supported version of CAVS.
> > > > > >
> > > > > > Please let me know what you prefer.
> > > > > >
> > > > > >
> > > > > > Thanks,
> > > > > > Olivier