summaryrefslogtreecommitdiff
path: root/source/a/pkgtools/scripts
diff options
context:
space:
mode:
authorPatrick J Volkerding <volkerdi@slackware.com>2018-09-06 06:15:46 +0000
committerEric Hameleers <alien@slackware.com>2018-09-06 18:00:33 +0200
commitce9883adcab118cd8a0ae45076486ebf643fe2d5 (patch)
tree5401dcca86cb602cf0a9b8443345458c51735cec /source/a/pkgtools/scripts
parent38dfc4aa131a87735dfc8629c3394c7d2670e9d5 (diff)
downloadcurrent-ce9883adcab118cd8a0ae45076486ebf643fe2d5.tar.gz
Thu Sep 6 06:15:46 UTC 201820180906061546
a/acpid-2.0.30-x86_64-1.txz: Upgraded. a/etc-15.0-x86_64-9.txz: Rebuilt. Added support for /etc/ld.so.conf.d/. Thanks to Qury. a/kernel-firmware-20180904_85c5d90-noarch-1.txz: Upgraded. a/kernel-generic-4.14.68-x86_64-1.txz: Upgraded. a/kernel-huge-4.14.68-x86_64-1.txz: Upgraded. a/kernel-modules-4.14.68-x86_64-1.txz: Upgraded. a/pkgtools-15.0-noarch-22.txz: Rebuilt. installpkg: prevent noise from the "stray cat" if tar hands it a broken pipe. Thanks to ivandi. Also (in the comments) provide a possibly better but untested solution. Thanks to SeB. a/sysvinit-2.90-x86_64-2.txz: Rebuilt. sysvinit-2.90 added an undocumented feature that tries to spawn agetty if it sees console= in the kernel command line. Avoiding the debate about whether this code belongs in init at all, at least such a change should be documented, and probably made opt-in. This patch disables the new behavior unless "sysvinit_agetty" is also seen in the kernel command line. It seems like this might have been something added for Hurd, and it might be wise to stick to what we've got rather than following upstream on this particular package. If anyone notices any other odd behavior, please let me know. Thanks to shastah. a/sysvinit-scripts-2.1-noarch-19.txz: Rebuilt. When checking for intel_pstate, direct the stderr to /dev/null in case the file doesn't exist in /sys. Thanks to ivandi. Allow forcing an option by editing /etc/default/cpufreq. ap/cups-filters-1.21.2-x86_64-1.txz: Upgraded. ap/ghostscript-9.24-x86_64-1.txz: Upgraded. Patched multiple -dSAFER sandbox bypass vulnerabilities. Thanks to Tavis Ormandy. For more information, see: https://www.ghostscript.com/doc/9.24/News.htm https://www.kb.cert.org/vuls/id/332928 (* Security fix *) ap/sudo-1.8.25-x86_64-1.txz: Upgraded. d/binutils-2.31.1-x86_64-2.txz: Rebuilt. elf.c (_bfd_elf_get_symbol_version_string): Return _("<corrupt>") for corrupt symbol version info. elflink.c (bfd_elf_record_link_assignment): Always clear h->verinfo.verdef when overriding a dynamic definition. Thanks to Michael Short for pointing out the upstream patches. d/ccache-3.4.3-x86_64-1.txz: Upgraded. d/kernel-headers-4.14.68-x86-1.txz: Upgraded. d/llvm-6.0.1-x86_64-2.txz: Rebuilt. Recompiled with -DCLANG_BUILD_SHARED_LIBS=ON to fix issues when multiple OpenCL drivers are installed. Thanks to Heinz Wiesinger. d/mercurial-4.7.1-x86_64-1.txz: Upgraded. k/kernel-source-4.14.68-noarch-1.txz: Upgraded. l/mozilla-nss-3.39-x86_64-1.txz: Upgraded. n/curl-7.61.1-x86_64-1.txz: Upgraded. This update fixes an NTLM password overflow via integer overflow. For more information, see: https://curl.haxx.se/docs/CVE-2018-14618.html https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2018-14618 (* Security fix *) n/gnupg2-2.2.10-x86_64-1.txz: Upgraded. n/libtirpc-1.1.4-x86_64-1.txz: Upgraded. n/netatalk-3.1.11-x86_64-2.txz: Rebuilt. rc.atalk: fixed errors in status output. Thanks to marav. n/nghttp2-1.33.0-x86_64-1.txz: Upgraded. n/p11-kit-0.23.14-x86_64-1.txz: Upgraded. n/stunnel-5.49-x86_64-1.txz: Upgraded. x/libdrm-2.4.94-x86_64-1.txz: Upgraded. x/xf86-video-ati-20180824_de88ea27-x86_64-1.txz: Upgraded. xap/hexchat-2.14.2-x86_64-1.txz: Upgraded. xap/mozilla-firefox-60.2.0esr-x86_64-1.txz: Upgraded. This release contains security fixes and improvements. For more information, see: https://www.mozilla.org/security/known-vulnerabilities/firefoxESR.html (* Security fix *) xap/xlockmore-5.56-x86_64-1.txz: Upgraded. isolinux/initrd.img: Rebuilt. kernels/*: Upgraded. usb-and-pxe-installers/usbboot.img: Rebuilt.
Diffstat (limited to 'source/a/pkgtools/scripts')
-rw-r--r--source/a/pkgtools/scripts/installpkg23
1 files changed, 22 insertions, 1 deletions
diff --git a/source/a/pkgtools/scripts/installpkg b/source/a/pkgtools/scripts/installpkg
index 0d82f89c..9e75ef03 100644
--- a/source/a/pkgtools/scripts/installpkg
+++ b/source/a/pkgtools/scripts/installpkg
@@ -493,7 +493,28 @@ for package in $* ; do
fi
# The stray cat reduces the frequency of the lack of reported size.
# If it still fails, we hit it with a bigger hammer down below.
- cat $package | $packagecompression -dc | LC_ALL=C dd 2> $TMP/tmpsize${MCOOKIE} | cat | tar tf - 2> /dev/null 1> $TMP/tmplist${MCOOKIE}
+ #
+ # SeB's proposed fix for this. Untested.
+ #cat $package | $packagecompression -dc | LC_ALL=C dd 2> $TMP/tmpsize${MCOOKIE} | ( tar tf - ; cat > /dev/null ) 2> /dev/null 1> $TMP/tmplist${MCOOKIE}
+ # Some background info from SeB's email:
+ # As you can see, when dd doesn't print anything, it returns 141,
+ # which means it has got a SIGPIPE. The problem is hence some tar
+ # archives cause tar to exit before having emptied stdin. That's
+ # why adding a cat before tar sometimes makes things better (tar
+ # causes it to get a SIGPIPE some ms before it causes itself dd to
+ # get one)
+ #
+ # A fix is then to replace:
+ #
+ # cat ... | $decompressor ... | ... dd ... | cat ... | tar tf - >...
+ #
+ # with:
+ #
+ # $decompressor ... | ... dd ... | ( tar tf -; cat >/dev/null ) >...
+ #
+ # This way the last cat will either exit because there's nothing on
+ # stdin or purge this one properly so dd gets the EOF it needs.
+ cat $package | $packagecompression -dc | LC_ALL=C dd 2> $TMP/tmpsize${MCOOKIE} | cat 2> /dev/null | tar tf - 2> /dev/null 1> $TMP/tmplist${MCOOKIE}
TARERROR=$?
if [ ! "$TARERROR" = "0" ]; then
EXITSTATUS=1 # tar file corrupt