• PHP2Shell

    From not@not@telling.you.invalid (Computer Nerd Kev) to comp.misc on Thu Oct 22 08:41:44 2026
    From Newsgroup: comp.misc

    In case I'm not the last to hear about it:

    "Developers of the PHP programming language have identified two
    critical vulnerabilities and have been urging website operators to
    run an update to fix the issues.

    The issue was first discovered using artificial intelligence,
    according to a Google-owned cybersecurity agency.

    PHP underpins large parts of the internet, raising concerns that a
    significant number of websites could be vulnerable to attack.

    Among those known to be affected are European government
    departments and popular sites like Wikipedia." ...

    https://www.abc.net.au/news/2026-10-22/php-vulnerabilities-cyber-asd/107294812

    https://www.php.net/archive/2026.php#2026-10-21-5
    --
    __ __
    #_ < |\| |< _#
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From noel@deletethis@invalid.lan to comp.misc on Sun Oct 25 09:51:30 2026
    From Newsgroup: comp.misc

    On Thu, 22 Oct 2026 08:41:44 +1000, Computer Nerd Kev wrote:

    In case I'm not the last to hear about it:

    "Developers of the PHP programming language have identified two
    critical vulnerabilities and have been urging website operators to run
    an update to fix the issues.

    The issue was first discovered using artificial intelligence, according
    to a Google-owned cybersecurity agency.

    PHP underpins large parts of the internet, raising concerns that a
    significant number of websites could be vulnerable to attack.

    Among those known to be affected are European government departments
    and popular sites like Wikipedia." ...

    and Australian Govt depts, including ASD.


    https://www.abc.net.au/news/2026-10-22/php-vulnerabilities-cyber-
    asd/107294812

    https://www.php.net/archive/2026.php#2026-10-21-5

    You are it seems :D Patches/updated sources released around midnight Wed local, I finished updates here (cause we use the source, luke) around 6am
    Thu local (due the torrential weather alarms, I had been up since 430
    anyway)

    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Computer Nerd Kev@not@telling.you.invalid to comp.misc on Sun Oct 25 11:24:02 2026
    From Newsgroup: comp.misc

    noel <deletethis@invalid.lan> wrote:
    You are it seems :D Patches/updated sources released around midnight Wed local, I finished updates here (cause we use the source, luke) around 6am Thu local (due the torrential weather alarms, I had been up since 430 anyway)

    Debian still haven't sorted out how to apply them though. The
    updated Debian 12 (Bookworm) and 13 (Trixie) PHP packages broke all
    sorts of things because of this bug that stops PHP modules loading
    (as I posted about in alt.os.linux.debian), which the maintainers
    sure are taking their time looking into:

    https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1151659

    Might be time for me to "use the source, Luke" too...
    --
    __ __
    #_ < |\| |< _#
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From noel@deletethis@invalid.lan to comp.misc on Sun Oct 25 20:24:55 2026
    From Newsgroup: comp.misc

    On Sun, 25 Oct 2026 11:24:02 +1000, Computer Nerd Kev wrote:

    noel <deletethis@invalid.lan> wrote:
    You are it seems :D Patches/updated sources released around midnight
    Wed local, I finished updates here (cause we use the source, luke)
    around 6am Thu local (due the torrential weather alarms, I had been up
    since 430 anyway)

    Debian still haven't sorted out how to apply them though. The updated
    Debian 12 (Bookworm) and 13 (Trixie) PHP packages broke all sorts of
    things because of this bug that stops PHP modules loading (as I posted
    about in alt.os.linux.debian), which the maintainers sure are taking
    their time looking into:

    https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1151659

    Might be time for me to "use the source, Luke" too...

    on the one debian box i have that stays true to their repo, no sources,
    is the freepbx 17 server, debian 12, I did not manualy update it, its
    daily apt run did that, much closer to Thur midnight early Friday


    if this helps... this is what is installed, sangoma are slack arses and
    dont use fast-cgi, just mod_php... but its far more memory efficient
    anyway than f/cgi which is I guess why they stuck with it.


    libapache2-mod-php8.2/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed,automatic]

    php-common/oldstable,oldstable,now 2:93 all [installed,automatic]

    php-pear/oldstable,oldstable,now 1:1.10.13+submodules+notgz+2022032202-2
    all [installed]

    php8.2-bcmath/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed]

    php8.2-bz2/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed]

    php8.2-cli/oldstable-security,now 8.2.35-1~deb12u1 amd64
    [installed,automatic]

    php8.2-common/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed,automatic]

    php8.2-curl/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed]

    php8.2-gd/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed]

    php8.2-igbinary/oldstable,oldstable,now 3.2.13-1 amd64
    [installed,automatic]

    php8.2-intl/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed]

    php8.2-ldap/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed]

    php8.2-mbstring/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed]

    php8.2-mysql/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed]

    php8.2-opcache/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed,automatic]

    php8.2-readline/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed,automatic]

    php8.2-redis/oldstable,oldstable,now 5.3.7+4.3.0-3 amd64 [installed]

    php8.2-soap/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed]

    php8.2-sqlite3/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed]

    php8.2-ssh2/oldstable,oldstable,now 1.3.1+0.13-7 amd64 [installed]

    php8.2-xml/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed]

    php8.2-zip/oldstable-security,now 8.2.35-1~deb12u1 amd64 [installed]

    php8.2/oldstable-security,now 8.2.35-1~deb12u1 all [installed]


    cheers
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Lawrence =?iso-8859-13?q?D=FFOliveiro?=@ldo@nz.invalid to comp.misc on Sun Oct 25 20:23:28 2026
    From Newsgroup: comp.misc

    On 25 Oct 2026 20:24:55 +1000, noel wrote:

    ... sangoma are slack arses and dont use fast-cgi, just mod_php...

    You mean PHP-FPM rather than FastCGI? PHP-FPM reimplements the entire (Apache-specific) mod_php environment within a standalone process,
    using FastCGI to communicate with the web server. This allows PHP code
    written to run with mod_php to work with other web servers, so long as
    they have FastCGI support.

    As far as I know, nobody else uses FastCGI for anything.

    ...but its far more memory efficient anyway than f/cgi which is I
    guess why they stuck with it.

    Interesting to know ... but the right way to do things is to run your
    code in its own process anyway, with its own mini-HTTP server to
    communicate with the rCLrealrCY web server via what they call a rCLreverserCY proxy. That works with any web server, is language-independent (so
    long as you have a decent basic HTTP server library to call on), and
    leaves you free to manage your own event loop, with support for
    arbitrary background tasks etc.
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From not@not@telling.you.invalid (Computer Nerd Kev) to comp.misc on Mon Oct 26 11:37:38 2026
    From Newsgroup: comp.misc

    noel <deletethis@invalid.lan> wrote:
    On Sun, 25 Oct 2026 11:24:02 +1000, Computer Nerd Kev wrote:
    Debian still haven't sorted out how to apply them though. The updated
    Debian 12 (Bookworm) and 13 (Trixie) PHP packages broke all sorts of
    things because of this bug that stops PHP modules loading (as I posted
    about in alt.os.linux.debian), which the maintainers sure are taking
    their time looking into:

    https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1151659

    Might be time for me to "use the source, Luke" too...

    on the one debian box i have that stays true to their repo, no sources,
    is the freepbx 17 server, debian 12, I did not manualy update it, its
    daily apt run did that, much closer to Thur midnight early Friday

    It turns out it's an Apache issue. The install script does a
    "restart" of Apache, but it reuses the old mod_php in memory
    instead of the newly installed one. But Apache does reload the PHP
    modules, and they all fail since they're for the new version.

    It's fixed by running "apache2ctl stop", then "apache2ctl start",
    instead of "apache2ctl restart". Or rebooting. Unfortunately Apache
    probably remains vulnerable to PHP2Shell until you do that.
    --
    __ __
    #_ < |\| |< _#
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From noel@deletethis@invalid.lan to comp.misc on Tue Oct 27 15:43:03 2026
    From Newsgroup: comp.misc

    On Mon, 26 Oct 2026 11:37:38 +1000, Computer Nerd Kev wrote:


    It's fixed by running "apache2ctl stop", then "apache2ctl start",
    instead of "apache2ctl restart". Or rebooting. Unfortunately Apache
    probably remains vulnerable to PHP2Shell until you do that.

    I would expect restart to restart for something like that as apache will
    not reload the newbie without it as you found out, i cant check reliably
    now, since freepbx had a few security updates that restarted the server,
    but that was 18hrs after /shrug/


    I see debian use their own custom apachectl, not the same as upstream but doesnt appear to do any harm at a quick look, just checked fpbx's php
    info page, says 8.2.35 so is running correct mod_php

    We are expecting another php update soon, so we check again in near
    future.

    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From noel@deletethis@invalid.lan to comp.misc on Tue Oct 27 15:48:48 2026
    From Newsgroup: comp.misc

    On Sun, 25 Oct 2026 20:23:28 +0000, Lawrence DrCOOliveiro wrote:

    On 25 Oct 2026 20:24:55 +1000, noel wrote:

    ... sangoma are slack arses and dont use fast-cgi, just mod_php...

    You mean PHP-FPM rather than FastCGI? PHP-FPM reimplements the entire (Apache-specific) mod_php environment within a standalone process,
    using FastCGI to communicate with the web server. This allows PHP code written to run with mod_php to work with other web servers, so long as
    they have FastCGI support.


    yes yes, i was distracte and used the wrong term, I know, on usenet thats cardinal sin, but like IGAF :)

    As far as I know, nobody else uses FastCGI for anything.

    ...but its far more memory efficient anyway than f/cgi which is I guess
    why they stuck with it.


    Interesting to know ... but the right way to do things is to run your
    code in its own process anyway

    Running a soho server might be fine, but if you try run that on shared
    hosting servers, you just opened a new DoS vector exhausting memory, so
    you weigh up the pros and cons
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Richard Kettlewell@invalid@invalid.invalid to comp.misc on Tue Oct 27 08:24:15 2026
    From Newsgroup: comp.misc

    noel <deletethis@invalid.lan> writes:
    Computer Nerd Kev wrote:

    It's fixed by running "apache2ctl stop", then "apache2ctl start",
    instead of "apache2ctl restart". Or rebooting. Unfortunately Apache
    probably remains vulnerable to PHP2Shell until you do that.

    I would expect restart to restart for something like that as apache will
    not reload the newbie without it as you found out, i cant check reliably now, since freepbx had a few security updates that restarted the server,
    but that was 18hrs after /shrug/

    I see debian use their own custom apachectl, not the same as upstream but doesnt appear to do any harm at a quick look, just checked fpbx's php
    info page, says 8.2.35 so is running correct mod_php

    DebianrCOs apachectl handles restart as follows:

    case "$ARGV" in
    restart|graceful)
    if $HTTPD ${APACHE_ARGUMENTS} -t 2> /dev/null ; then
    $HTTPD ${APACHE_ARGUMENTS} -k "$ARGV"
    else
    $HTTPD ${APACHE_ARGUMENTS} -t
    fi
    ERROR=$?
    ;;

    i.e. it syntax checks the configuration and then runs apache2 -k restart.

    https://httpd.apache.org/docs/current/stopping.html#hup describes the
    effect of this option:

    apachectl -k restart

    Sending the HUP or restart signal to the parent causes it to kill
    off its children like in TERM, but the parent doesn't exit. It
    re-reads its configuration files, and re-opens any log files. Then
    it spawns a new set of children and continues serving hits.

    i.e. in Apache speak, rCLrestartrCY doesnrCOt actually mean restarting the server at all, at least in the common sense of restart meaning stop and
    then start from scratch.

    What that means for PHP configuration lingering past one of this fake
    restarts I couldnrCOt say, but it does make the behavior CNK reports
    completely unsurprising.

    systemctl restart should actually restart the server.
    --
    https://www.greenend.org.uk/rjk/
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From not@not@telling.you.invalid (Computer Nerd Kev) to comp.misc on Wed Oct 28 08:28:39 2026
    From Newsgroup: comp.misc

    Richard Kettlewell <invalid@invalid.invalid> wrote:
    https://httpd.apache.org/docs/current/stopping.html#hup describes the
    effect of this option:

    apachectl -k restart

    Sending the HUP or restart signal to the parent causes it to kill
    off its children like in TERM, but the parent doesn't exit. It
    re-reads its configuration files, and re-opens any log files. Then
    it spawns a new set of children and continues serving hits.

    I see, I checked the apache2ctl man page's description on Debian
    but it didn't bother to mention "the parent doesn't exit":

    restart Restarts the Apache daemon by sending it a SIGHUP. If
    the daemon is not running, it is started. This command
    automatically checks the configuration files via con-
    figtest before initiating the restart to to catch the
    most obvious errors. However, it is still possible for
    the daemon to die because of problems with the configura-
    tion.

    i.e. in Apache speak, "restart" doesn't actually mean restarting the
    server at all, at least in the common sense of restart meaning stop and
    then start from scratch.

    Yeah, I guess I didn't know to read the man page in Apache speak.
    In my case Lograte does a "graceful restart" whenever the log file
    is changed, so it's guaranteed to go wrong at least shortly after
    the update. Which also means this can't have been a problem with
    earlier PHP updates. I'm not sure why it wasn't stopped and started automatically when the update ran. The package's scripts are
    supposed to run "invoke-rc.d apache2 restart" after installation,
    which the Apache logs show causes it to do a "graceful stop" after
    receiving SIGWINCH (that "restart" isn't in Apache speak) when
    run manually.
    --
    __ __
    #_ < |\| |< _#
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Lawrence =?iso-8859-13?q?D=FFOliveiro?=@ldo@nz.invalid to comp.misc on Tue Oct 27 22:36:52 2026
    From Newsgroup: comp.misc

    On 27 Oct 2026 15:48:48 +1000, noel wrote:

    On Sun, 25 Oct 2026 20:23:28 +0000, Lawrence DrCOOliveiro wrote:

    ... the right way to do things is to run your code in its own
    process anyway

    Running a soho server might be fine, but if you try run that on
    shared hosting servers, you just opened a new DoS vector exhausting
    memory ...

    The Web server runs in its own process(es); either it is also rCLa new
    DoS vector exhausting memoryrCY, or it has ways of managing nuisance
    requests. Which I can also use in my own code.

    And my application server will be behind the Web server, connected via
    a rCLreverserCY proxy (donrCOt know why they call it that), for added protection. And also to let the full-strength Web server handle all
    the complicated stuff, like TLS and virtual hosting.
    --- Synchronet 3.22a-Linux NewsLink 1.2