[PATCH 1/1] wininet/test: Check null pointer inInternetGetSecurityInfoByURLW

[Wei Xie]

Hi Jacek:

I don't received your modified patch.

          CertFreeCertificateChain(chain);
+
+        res = pInternetGetSecurityInfoByURLA(url, &chain, NULL);
+        ok_(__FILE__,line)(res, "InternetGetSecurityInfoByURLA failed: %u\n", GetLastError());
+
+        res = pInternetGetSecurityInfoByURLA(url, NULL, &flags);
+        ok_(__FILE__,line)(res, "InternetGetSecurityInfoByURLA failed: %u\n", GetLastError());
+
+        SetLastError(0xdeadbeef);
+        res = pInternetGetSecurityInfoByURLA(url, NULL, NULL);
+        ok_(__FILE__,line)(!res && GetLastError() == ERROR_INVALID_PARAMETER,
+                           "InternetGetSecurityInfoByURLA returned: %x(%u), expected failed %u\n",
+                            res, GetLastError(), ERROR_INVALID_PARAMETER);
     }else {

This is my modified version, is this ok? Thanks for your help.

------------------ Original ------------------
From:  "Jacek Caban"<jacek at codeweavers.com>;
Date:  Tue, Jan 9, 2018 07:09 PM
To:  "Jactry Zeng"<jactry92 at gmail.com>; "Wei Xie"<xiewei at deepin.com>;
Cc:  "wine-devel"<wine-devel at winehq.org>;
Subject:  Re: [PATCH 1/1] wininet/test: Check null pointer inInternetGetSecurityInfoByURLW
 
On 09.01.2018 09:28, Jactry Zeng wrote:
Hi Wei,

Thanks for contribution!
I have a question about the tests:

+
+        SetLastError(0);
+        res = pInternetGetSecurityInfoByURLA(url, &chain, NULL);
+        ok_(__FILE__,line)(res && GetLastError() == 0,
+                           "InternetGetSecurityInfoByURLA failed returned: %x(%u), expected success\n", res, GetLastError());
+
+        SetLastError(0);
+        res = pInternetGetSecurityInfoByURLA(url, NULL, &flags);
+        ok_(__FILE__,line)(res && GetLastError() == 0,
+                           "InternetGetSecurityInfoByURLA failed returned: %x(%u), expected success\n", res, GetLastError());
     }else {

Why not set last error as 0xdeadbeef in these two tests?

Actually (with a few exceptions when apps really depend on it), it's better to not test last error on success. Caller should never use it anyway and it might be polluted by internal calls that we're not interested in.


Wei, I sent a modified version of your patch. I hope you're fine with those changes. Thanks for contribution.


Jacek