Hey MSE, why did you make me panic? It told Apache files are Trojan:Win32/Critet.BS.

Update (Mar.22):

   Today, I’ve updated Apache from 2.4.32 to 2.4.33. Now, MSE says all files are clean. What was that alert? Really compromised or not? Anyway, I backed the MSE settings to the default.

   This morning Microsoft Security Essentials suddenly told Apache files are Trojan:Win32/Critet.BS and quarantined them, so Apache stopped on my server PC whose OS is Windows7 HE SP1. Although I needed to recover the service immediately, I had to take my mother to a hospital. Therefore the service must have been unavailable for about an hour.
Continue reading “Hey MSE, why did you make me panic? It told Apache files are Trojan:Win32/Critet.BS.”

Search Console gave me “Security issues” again.

   Last night, I found Search Console gave me "Security issues(1)(2)(3)" again, when I logged on its HOME. This was the second times. Last time was on May 9.

   Both of them are not by something like malware but by my fault. orz Continue reading “Search Console gave me “Security issues” again.”

Attacks against WordPress 4.7 or 4.7.1 are still increasing.

   Still increasing. Do updating you WordPress ASAP. I read some articles and also got an email about this from my ISP. So I’m writing this.

   I talked about this with くりくりさん on Twitter. I first mentioned it on February 6. It was because of my finding a 徳丸’s post, “WordPress 4.7.1 の権限昇格脆弱性について検証した”.

   Yesterday, Security Next told us some IP addresses about attackers. I checked up on my log last night. I found an access from one of questionable IP addresses, which was on February 6. It caused 500 error on my server. Maybe because my WordPress was already version 4.7.2 at this point.
   Its user-agent is python-requests/2.11.1 and its destination is /wp-json/wp/v2/posts/.

   WordPress 4.7.2 was released more than a week ago, and WordPress has an auto-update feature enabled by default, along with an easy manual update process. Despite this, this situation. It’s indeed disappointing.

I’ve updated to phpMyAdmin4.6.6.

   I’ve updated to phpMyAdmin4.6.6. After that, the new version gave me “OpenSSL error: error:0607A082:digital envelope routines:EVP_CIPHER_CTX_set_key_length:invalid key length” at HOME when I logged in.
   This is maybe because of this 👉 $cfg[‘Servers’][$i][‘ssl_verify’].

   The page says “Disabling the certificate verification defeats purpose of using SSL. This will make the connection vulnerable to man in the middle attacks.”, but my SQL server and phpMyAdmin don’t accept accesses from outside of NAT router and the user is only me. So, as my temporal workaround, I added the next line to my config.inc.php.

$cfg['Servers'][$i]['ssl_verify'] = false;

Windows10 1511 sneaking into my CF-J10 during last night.

Update information      Edit(Nov.28)  Edit2(2016.Aug.10)

   Windows10 1511 was sneaking into my CF-J10 during last night because it was a normal Windows Update and I left CF-J10 hibernating instead I made it shutdown and unpluged completely.

   Fortunately, I have no fatal problems. But I have something as below. Continue reading “Windows10 1511 sneaking into my CF-J10 during last night.”

Updating Apache because of CVE-2015-1793.

   I updated my Apache 2.4.12(httpd-2.4.12-win32-VC14.zip) to 2015 Jul 9 version because of Alternative chains certificate forgery (CVE-2015-1793).

   It is built with ‘IPv6 Crypto apr-1.5.1 apr-util-1.5.4 apr-iconv-1.2.1 openssl-1.0.2d zlib-1.2.8 pcre-8.37 libxml2-2.9.2 lua-5.1.5 expat-2.1.0′. Its Changelog.
   This version is also built with the latest Windows® Visual Studio C++ 2015 RC aka VC14. I began to use VC14 version on June 2 due to OpenSSL 1.0.2. If you use the version, you need to install vc_redist_x64/86.exe before installing the version.

   I really appreciate Steffen’s hard and quick work. Thanks again, Steffen.

   By the way, I take this occasion to update to phpMyAdmin 4.4.11 and MariaDB 10.0.20.

   About phpMyAdmin I noticed two differences. From the version 4.4.10 the download URL changed from sourceforge.net to phpmyadmin.net. And this version, I mean 4.4.11, they provide not only MD5/SHA1 but PGP. I wonder if something happened between sourceforge and phpmyadmin.

Yesterday, FireFox 39.0 came.

   Yesterday, FireFox 39.0 came by automatic update.

   Now FireFox deploys fixes for the Logjam attack really. You can see what vulnerabilities are fixed in 39.0 ⇒ Fixed in Firefox 39.

   As you know, they fix a lot of vulnerabilities in each version. So you must keep your web browser up-to-date status. Well, this is not for web browsers only (^^;).

Although belated, about Logjam.

Update information      Edit   Edit2(Jul.7)   Edit3(Sep.2)

   Yesterday, I came home around 8 pm and saw the first fireflies of this year in my garden. Wow!

Server Test1   By the way, I read the article “TLSに脆弱性「Logjam」 – 国家レベルなら1024ビットまで盗聴可能” on May 21. Then I went to Guide to Deploying Diffie-Hellman for TLS and did Server Test. I got the result like the right image. Before the test, despite I didn’t do anything else more than I had done until 2014.Oct.28 (= A self-sighed certificate with SANs and SHA256 by OpenSSL).

   And that night, I had a comment from くりくりさん on my Japanese blog. He let me know about Logjam. I wrote back him that I tried writing about Logjam and I’m writing it now, ha-ha.

   When I tested my server at the first time, the server supported the following Cipher Suites.


   But actually I don’t need most of them. Because the user of my SSL server is only me and I usually use the latest version Web Browsers as I always say. I only use ECDHE-RSA-AES128-GCM-SHA256 suite at my access. So I changed SSLCipherSuite directive on my ssl.conf like this.
   SSLCipherSuite ECDHE-RSA-AES128-GCM-SHA256
   This configuration is not useful for other servers. If you want to know a realistic configuration, see Guide to Deploying Diffie-Hellman for TLS. If your server is in newer versions of Apache (2.4.8 and newer) and OpenSSL 1.0.2 or later, you can directly specify your DH params file. But even if your server isn’t, you can use SSLCipherSuite and SSLProtocol instead of SSLOpenSSLConfCmd and can make your server safe from Logjam attack.

Sever Test2   Actually, ApacheLounge version HTTPD is still built with OpenSSL 1.0.1 branch. So I could not use SSLOpenSSLConfCmd directive. But after changing my SSLCipherSuite, I got the result like right image.

Another Test   Another Logjam Attack Checker gave me the right result.

   In addition, when using Apache 2.4 with OpenSSL 1.0.1 and later, SSLProtocol all means +SSLv3 +TLSv1 +TLSv1.1 +TLSv1.2. When using Apache 2.4.7 or later, aNULL, eNULL and EXP ciphers are always disabled.

   According to the The Logjam Attack page, Google Chrome (including Android Browser), Mozilla Firefox, Microsoft Internet Explorer, and Apple Safari are all deploying fixes for the Logjam attack. But still now (9:45 am JST), I have Warning! Your web browser is vulnerable to Logjam and can be tricked into using weak encryption. You should update your browser when I access the page by FireFox 38.0.1, Google Chrome 43.0.2357.65 or SeaMonkey 2.33.1. Only Internet Explorer 11.0.19 gives Good News! Your browser is safe against the Logjam attack.
Note) I don’t check this with other browsers and versions.

   Yesterday, FireFox 39.0 came. Now I have Good News! Your browser is safe against the Logjam attack by it.

   I’ve not checked it for a while. Today, Google Chrome ver. 45.0.2454.85 has come, so I check just now. The site gives Good News! Your browser is safe against the Logjam attack. When was Chrome deploying fixes for it? I have no idea!!

   Now 1:00a.m. SeaMonkey’s new version 2.35 has come after long interval. And, I’ve finally had Good News! Your browser is safe against the Logjam attack by it.

A self-sighed certificate with SANs and SHA256 by OpenSSL.

Update information      Edit(Oct.28)

   When I tested my SSL server by “Qualys SSL Labs – Projects / SSL Server Test” for this dust, the test gave me following Reds and Oranges (^_^;).

  1. Trusted : No NOT TRUSTED <<---- Because I use a self-sighed certificate that the Labs doesn't know. So I ignore the message with confidence ha-ha.
  2. IE 6 / XP No FS 1 No SNI 2 : Protocol or cipher suite mismatch : Fail3 <<---- My SSL server user is only me, and I don't use IE 6 / XP. So I ignore the message.
  3. Fail3 They say “Only first connection attempt simulated. Browsers tend to retry with a lower protocol version.” My SSL server doesn’t accept lower protocols, but it’s no problem for me.
  4.    As above I have nothing to be done about Reds.


  1. Prefix handling : Not valid for “www.o6asan.com” :CONFUSING
  2. Signature algorithm : SHA1withRSA : WEAK
  3. Chain issues : Contains anchor <<---- Ivan Ristić replied about “Chain issues Contains anchor”. So I ignore the message.
  4. Not in trust store <<---- Because I use a self-sighed certificate. So I ignore the message.
  5. Downgrade attack prevention : No, TLS_FALLBACK_SCSV not supported
  6. Forward Secrecy : With some browsers

   As above I have something to be done about 1, 2, 5 and 6. First I handle 5 and 6 because I need to re-create a new certificate for 1 and 2.

  1. I update Apache 2.4.10 (httpd-2.4.10-win32-VC11.zip) to October 20 version. Because it was built with openssl-1.0.1j which supported TLS_FALLBACK_SCSV.
  2. I uncommented SSLHonorCipherOrder on and changed SSLCipherSuite Directive value in the httpd-ssl.conf.

    !PSK !SRP !DSS

       Ref : Configuring Apache, Nginx, and OpenSSL for Forward Secrecy
    ↓ I changed on Dec. 23 because of RC4.

       Ref : RC4 in TLS is Broken: Now What?

    IF your server should support some mobile OS/browser or legacy ones, you need more additional configuration. But the above is enough to my server.
  3. After them the test gives me “Downgrade attack prevention : Yes, TLS_FALLBACK_SCSV supported” and “Forward Secrecy : Yes (with most browsers) ROBUST”.

   Second I handle 1 and 2.
   The 1 is caused by my self-sighed certificate only has o6asan.com as its Common Name. So I have to create a new certificate supporting both o6asan.com and www.o6asan.com. But there is a problem that I want to use only one IP address for my SSL server. Nowadays we have the solution for this issue that is called SNI(Server Name Indication) though all OS/browser haven’t supported it yet. I can use a wildcard certificate or SAN for SNI. I use SANs because I don’t want to allow my SSL server to accept all sub domains though I can restrict them by Apache configuration.
   The 2 is caused by my creation of the certificate by OpenSSL default, which is set to use SHA1. So I’ll use default_md = sha256 for the new certificate.
 On 28th, I re-read Server Name Indication. Is SNI a different story from wildcard certificate / SAN? I don’t understand them still now. Difficult.

   I copy the file openssl.cnf(← this is the default name) from Apche24\conf folder to c:\openssl-1.0.1x-winxx\ssl(← this is the default location) and customize like the followings.

    Change values and uncomment a line.

  1. dir = ./demoCA —->> dir = X:/demoCA <<----Need an exact path
  2. default_crl_days = 30 —->> default_crl_days = 365
  3. default_md = default —->> default_md = sha256
  4. default_bits = 1024 —->> default_bits = 2048
  5. # req_extensions = v3_req —->> req_extensions = v3_req
    Adding lines.

  1. subjectAltName = @alt_names to [ v3_req ] area.
  2. [ alt_names ]
    DNS.1 = example.com
    DNS.2 = www.example.com
    to just before [ v3_ca ] area.
    You can add your domains, like DNS.1, DNS.2, DNS.3, ….
  3. If you make a client certificate, add the followings to the end of the openssl.cnf.
    [ ssl_client ]
    basicConstraints = CA:FALSE
    nsCertType = client
    keyUsage = digitalSignature, keyEncipherment
    extendedKeyUsage = clientAuth
    nsComment = "OpenSSL Certificate for SSL Client"

   Now I’ll create new certificate. (Ref : WordPress: Administration Over SSL #1)

    ||Create myCA||

  1. Make myCA folder at X:/
  2. Make two folders and a file named private, newcerts and index.txt in the myCA.
  3. Run cmd.exe as Administrator
    pushd X:\myCA
    echo 01 > serial
    openssl req -new -keyout private\cakey.pem -out careq.pem
    openssl ca -selfsign -in careq.pem -extensions v3_ca -out cacert.pem
    copy cacert.pem (Drive_SV):\Apache24\conf\ssl.crt
    copy cacert.pem my_ca.crt

      Note) (Drive_SV) is a partition for server components on my home server PC.
    ||Create Server Cert||

  1. pushd X:\myCA
    openssl genrsa -out server.key 2048
    openssl req -new -out server.csr -key server.key
  2. Check multiple SANs in the CSR (Can you see ‘Subject Alternative Name’ area in it?)
    openssl req -text -noout -in server.csr
  3. openssl ca -in server.csr -out server.crt -extensions v3_req
    copy server.key cp_server.key
    openssl rsa <cp_server.key> server.key
    copy server.key (Drive_SV):\Apache24\conf
    copy server.crt (Drive_SV):\Apache24\conf
    ||Create Client Cert||

  1. pushd X:\myCA
    openssl req -new -keyout client.key -out client.csr
    openssl ca -policy policy_anything -extensions ssl_client -in client.csr -out client.crt
    openssl pkcs12 -export -in client.crt -inkey client.key -out clientcert.p12

References about SANs : FAQ/subjectAltName (SAN), Multiple Names on One Certificate.

   I have a self-sighed certificate with SANs and SHA256 now. Mission complete!!

Does cURL have POODLE?

Update information      Edit(Oct.26)

   I wrote about “POODLE” issue on the last post. After that, I suddenly got worried about cURL on WordPress because I read SSLv3 fallback attack POODLE.

   Though I found a following option at curl_setopt,
I couldn’t get where I should add it among WordPress Core Scripts. So, I made a topic on WordPress Forums…I’m waiting answers.

   I just made the topic [resolved]. Because I got the result that my cURL exactly uses TLSv1.2 by %{SSL_PROTOCOL} on the Apache log. I don’t need CURL_SSLVERSION_TLSv1 on the file class-http.php. If the SSL sever has appropriate configurations, clients can access it safely if their software components have the abilities required.

   Clap clap, (*´▽`*).