Vulnerability

之前没有特别注意这个漏洞，这里稍微记一笔。

PostgreSQL 包含一系列系统视图，这些系统视图可以用来查询系统表。 由于 pg_stats_ext 和 pg_stats_ext_exprs 这两个视图在 PostgreSQL 14-16 的 16.3、15.7 和 14.12 之前的版本中缺少了必要的访问控制， 因此未经授权的用户将可以通过这些视图访问其他用户通过 CREATE STATISTICS 创建的一系列统计数据，而这些数据可能会揭示这些未经授权的用户原本没有权限访问的数据，或是函数的执行结果。

PostgreSQL 在 16.3、15.7 和 14.12 中修正了这一问题，但这仅限于新安装的情况。 对于已经安装好的正在运行的数据库，管理员还必须重建这两个系统视图以收紧权限（加入了 WITH (security_barrier) 以及 WHERE 子句来限制访问）。 修正脚本位于源代码的 src/backend/catalog/fix-CVE-2024-4317.sql (对于 FreeBSD pkg 用户，这些修正脚本会安装到 /usr/local/share/postgresql/fix-CVE-2024-4317.sql)， 使用 psql 运行即可。

最近没怎么关注安全方面的进展，结果错过了去年年底披露的 SMTP Smuggling。 这是 Timo Longin 发现的一个全新的针对 SMTP 协议的攻击手法， 现在的年轻人真是蛮厉害的。

今天早上4点多就爬起来准备发安全公告（友商约的是早上5点发），一刷，OpenSSL官方已经正式发表了（时间大约是4点30），于是立即开始相关手续，commit修正，发公告，通知CERT，这些不表。

这次休假前头一天晚上收到了 ISC 发来的安全公告 CVE-2011-1910。我起草完 FreeBSD 安全公告 FreeBSD-SA-11:02.bind.asc 之后就去休假了，这里补上这个漏洞的说明。

无论你是否使用 DNSsec，只要用 BIND 9.x 来做 DNS 缓存解析服务（而不仅仅做Authoritive DNS服务），就有可能受到这个漏洞的影响。而攻击的方法也相当简单，攻击者只要建立一个 DNS 域，令在回应域名不存在的同时发回一个大的 RRSIG RRset，即可触发 BIND 的某处断言令其退出。

今天 Colin 在blog上介绍说 他刚刚修正了一个 TarSnap 重大安全漏洞。简单地说，用来加密密钥文件的AES-CTR实现时，计数器没有增加（正确的实现中计数器应该每次递增），导致在已知明文时能够通过加密块推算出一块数据使用的密钥，并用它来解密余下的数据。

作者 Bruce Schneier
原文 Debating Full Disclosure

作者授权 blog.delphij.net 发表此翻译版本。未经许可请勿转载。

全面披露 —— 将安全漏洞的细节公之于众 —— 是一种很好的做法。公开并让更多的人检查是改善安全唯一可靠的途径，而遮遮掩掩则只会让我们的安全受到损害。

最近在CU论坛的历险记。

讨论最开始是由有位大哥求助如何在FreeBSD中移去sendmail引发的。嗯，这将导致不幸，提醒一下。

接下来老大说sendmail是系统自带的，不像Solaris或Linux那样用包安装，但他要弄掉，弄得很干净。

呃……显然啊，FreeBSD不是一个组装的玩意，它提供了很多内建的东西，不过对于那些热爱自己DIY的人还是应该鼓励一下，于是回帖说用NO_SENDMAIL阻止构建，然后installworld之后删除没有碰过的文件。最后出于好意提醒了一句，「默认安装的留在那不会给你带来很大的损失，但删除可能会带来意想不到的后果。」

我们来看看那个人说的这段话：

I’d be really surprised if somebody is actually able to get a real-world attack on a real-world pgp key usage or similar out of it (and as to the covert channel, nobody cares). It’s a fairly interesting approach, but it’s certainly neither new nor HT-specific, or necessarily seem all that worrying in real life. (HT and modern CPU speeds just means that the covert channel is _faster_ than it has been before, since you can test the L1 at core speeds. I doubt it helps the key attack much, though, since faster in that case cuts both ways: the speed of testing the cache eviction may have gone up, but so has the speed of the operation you’re trying to follow, and you’d likely have a really hard time trying to catch things in real life).

It does show that if you want to hide key operations, you want to be careful. I don’t think HT is at fault per se.

Every quarter we got the same news: Linux did it again!

Yes, they DID it again. Many sites became victim during the last scan of awstats vulnerability, the most famous ones are www.phpbb.com, and moto.debian.org.tw, etc. You can Google the cracker organization and find more.

Why Linux is again and again vulnerable to these hack attempts? Why other systems doesn’t have such serious security issues even when an exploit is published? The answer is apparant: Linux did worst ever, among all Operating Systems, even when you include Windows. Imagine, a kernel which can permit normal users to gain root privilege.

Researchers have announced preliminary indications of previously unknown vulnerabilities in popular security algorithms that could permit hackers to easily install undetectable back doors into computer code or to counterfeit electronic signatures. French computer scientist Antoine Joux reported on Aug. 12 his discovery of a flaw in the MD5 algorithm, which is often used with digital signatures. The algorithm is known as a hash function, which extrapolates from all input a unique fingerprint; however, if a hacker could produce the same fingerprint with a different input stream, then the resulting hash collision would authenticate software as safe to download and execute even though it contains a back door. MD5 is employed by the open-source Apache Web server product as well as Sun Microsystems’ Solaris Fingerprint Database, and the flaw Joux uncovered means that a hacker can produce one hash collision in a few hours on a standard PC. Meanwhile, four Chinese researchers issued a paper reporting that the SHA-0 Secure Hash Algorithm could be subverted, while Israel Institute of Technology researchers Eli Biham and Rafi Chen revealed at the Crypto 2000 conference on Aug. 17 that they were investigating possible flaws in the SHA-1 algorithm, the only signing algorithm approved for use in the U.S. Digital Signature Standard. SHA-1, which is incorporated into popular programs such as SSL and PGP, is thought to be secure because knowingly producing hash collisions via existing methods is impossible. SHA-1 depends on a computer executing a routine 80 times as it tries to create a unique fingerprint, and Biham declared that he was able to copy the fingerprint for 36 of those 80 executions. If SHA-1 shares similar vulnerabilities with SHA-0, then attempts to falsify a fingerprint would be sped up about 500-fold.