Anaconda-devel December 2008

anaconda-devel@lists.fedoraproject.org

25 participants
100 discussions

[PATCH] Fix the 'ks' boot argument (#471812)
by David Cantrell 16 Dec '08

16 Dec '08

If you boot with just 'ks', loader will assume the kickstart file is on NFS and determine the path to it using the nextserver and bootfile DHCP options. I neglected to get around to fixing this for the big NM changeover. Talking to NetworkManager to get the nextserver and bootfile options is a bit complex now. There is no direct way to get to the Dhcp4Config object. You have to get ActiveConnections, iterate, get Devices, iterate, get Hal path, iterate, get Interface property, match interface name, break, get Dhcp4Config property, recurse, get Options dict, recurse, recurse, recurse, iterate, get nextserver and bootfile, combine to form 'url'. --- loader/nfsinstall.c | 178 ++++++++++++++++++++++++++++++++++++++++++++++----- 1 files changed, 162 insertions(+), 16 deletions(-) diff --git a/loader/nfsinstall.c b/loader/nfsinstall.c index 9a61cad..b5d4353 100644 --- a/loader/nfsinstall.c +++ b/loader/nfsinstall.c @@ -35,6 +35,7 @@ #include <sys/socket.h> #include <arpa/inet.h> #include <netdb.h> +#include <NetworkManager.h> #include "copy.h" #include "loader.h" @@ -369,7 +370,17 @@ void setKickstartNfs(struct loaderData_s * loaderData, int argc, int getFileFromNfs(char * url, char * dest, struct loaderData_s * loaderData) { char * host = NULL, *path = NULL, * file = NULL, * opts = NULL; char * chk = NULL, *ip = NULL; - int failed = 0; + int failed = 0, found = 0, type = 0; + char *interface = NULL; + char *key = NULL; + char *val = NULL; + char nextserver[INET_ADDRSTRLEN+1]; + char *bootfile = NULL; + char *dpath[3]; + struct in_addr addr; + struct hostent *hostent; + DBusMessage *reply[4]; + DBusMessageIter iter[13]; iface_t iface; if (kickstartNetworkUp(loaderData, &iface)) { @@ -380,39 +391,174 @@ int getFileFromNfs(char * url, char * dest, struct loaderData_s * loaderData) { /* if they just did 'linux ks', they want us to figure it out from * the dhcp/bootp information */ -/* XXX: fixme NetworkManager integration */ -/* if (url == NULL) { - char ret[47]; - ip_addr_t *tip; + /* get the IP4Config object for the active interface */ - if (!(netCfg.dev.set & PUMP_INTFINFO_HAS_NEXTSERVER)) { - logMessage(ERROR, "no bootserver was found"); + reply[0] = get_nm_dbus_property(NM_DBUS_INTERFACE, + NM_DBUS_PATH, "ActiveConnections"); + if (!reply[0]) { return 1; } - tip = &(netCfg.dev.nextServer); - if (!(netCfg.dev.set & PUMP_INTFINFO_HAS_BOOTFILE)) { - inet_ntop(tip->sa_family, IP_ADDR(tip), ret, IP_STRLEN(tip)); + dbus_message_iter_init(reply[0], &iter[0]); + dbus_message_iter_recurse(&iter[0], &iter[1]); + + while (dbus_message_iter_get_arg_type(&iter[1]) != + DBUS_TYPE_INVALID) { + dbus_message_iter_recurse(&iter[1], &iter[2]); + + while (dbus_message_iter_get_arg_type(&iter[2]) != + DBUS_TYPE_INVALID) { + dbus_message_iter_get_basic(&iter[2], &dpath[0]); + + reply[1] = get_nm_dbus_property(NM_DBUS_INTERFACE, + dpath[0], "Devices"); + if (!reply[1]) { + return 1; + } + + dbus_message_iter_init(reply[1], &iter[3]); + dbus_message_iter_recurse(&iter[3], &iter[4]); + + while (dbus_message_iter_get_arg_type(&iter[4]) != + DBUS_TYPE_INVALID) { + dbus_message_iter_recurse(&iter[4], &iter[5]); + + while (dbus_message_iter_get_arg_type(&iter[5]) != + DBUS_TYPE_INVALID) { + dbus_message_iter_get_basic(&iter[5], &dpath[1]); + + reply[2] = get_nm_dbus_property(NM_DBUS_INTERFACE, + dpath[1], "Interface"); + if (!reply[2]) { + return 1; + } + + dbus_message_iter_init(reply[2], &iter[6]); + dbus_message_iter_recurse(&iter[6], + &iter[7]); + + while (dbus_message_iter_get_arg_type(&iter[7]) + != DBUS_TYPE_INVALID) { + dbus_message_iter_get_basic(&iter[7], + &interface); - if (asprintf(&url, "%s:%s", ret, "/kickstart/") == -1) { + if (!strcmp(interface, iface.device)) { + found = 1; + break; + } + + dbus_message_iter_next(&iter[7]); + } + + if (found) break; + dbus_message_iter_next(&iter[5]); + } + + if (found) break; + dbus_message_iter_next(&iter[4]); + } + + if (found) break; + dbus_message_iter_next(&iter[2]); + } + + if (found) break; + dbus_message_iter_next(&iter[1]); + } + + if (!found) { + logMessage(ERROR, "%s has no BOOTP values\n", iface.device); + return 1; + } + + /* get Dhcp4Config object for this device */ + reply[2] = get_nm_dbus_property(NM_DBUS_INTERFACE, + dpath[1], "Dhcp4Config"); + if (!reply[2]) { + return 1; + } + + dbus_message_iter_init(reply[2], &iter[6]); + dbus_message_iter_recurse(&iter[6], &iter[7]); + + while (dbus_message_iter_get_arg_type(&iter[7]) != + DBUS_TYPE_INVALID) { + dbus_message_iter_get_basic(&iter[7], &dpath[2]); + + reply[3] = get_nm_dbus_property(NM_DBUS_INTERFACE, + dpath[2], "Options"); + if (!reply[3]) { + return 1; + } + + dbus_message_iter_init(reply[3], &iter[8]); + dbus_message_iter_recurse(&iter[8], &iter[9]); + + if (dbus_message_iter_get_arg_type(&iter[9]) != + DBUS_TYPE_ARRAY) { + dbus_message_iter_next(&iter[7]); + continue; + } + + dbus_message_iter_recurse(&iter[9], &iter[10]); + while (dbus_message_iter_get_arg_type(&iter[10]) == + DBUS_TYPE_DICT_ENTRY) { + dbus_message_iter_recurse(&iter[10], &iter[11]); + key = NULL; + dbus_message_iter_get_basic(&iter[11], &key); + dbus_message_iter_next(&iter[11]); + dbus_message_iter_recurse(&iter[11], &iter[12]); + type = dbus_message_iter_get_arg_type(&iter[12]); + dbus_message_iter_get_basic(&iter[12], &val); + + if ((!strcasecmp(key, "server_name")) && + (type == DBUS_TYPE_STRING)) { + memset(&nextserver, '\0', sizeof(nextserver)); + if (inet_pton(AF_INET, val, &addr) >= 1) { + strcpy(nextserver, val); + } else { + hostent = gethostbyname(val); + if (hostent != NULL) { + if (inet_ntop(AF_INET, hostent->h_addr_list[0], + nextserver, + INET_ADDRSTRLEN) == NULL) { + memset(&nextserver, '\0', sizeof(nextserver)); + } + } + } + } else if ((!strcasecmp(key, "filename")) && + (type == DBUS_TYPE_STRING)) { + bootfile = strdup(val); + } + + dbus_message_iter_next(&iter[10]); + } + + dbus_message_iter_next(&iter[7]); + } + + if (nextserver == NULL) { + logMessage(ERROR, "no bootserver was found"); + return 1; + } + + if (bootfile == NULL) { + if (asprintf(&url, "%s:/kickstart/", nextserver) == -1) { logMessage(CRITICAL, "%s: %d: %m", __func__, __LINE__); abort(); } logMessage(ERROR, "bootp: no bootfile received"); } else { - inet_ntop(tip->sa_family, IP_ADDR(tip), ret, IP_STRLEN(tip)); - - if (asprintf(&url, "%s:%s", ret, netCfg.dev.bootFile) == -1) { + if (asprintf(&url, "%s:%s", nextserver, bootfile) == -1) { logMessage(CRITICAL, "%s: %d: %m", __func__, __LINE__); abort(); } - logMessage(INFO, "bootp: bootfile is %s", netCfg.dev.bootFile); + logMessage(INFO, "bootp: bootfile is %s", bootfile); } } -*/ /* get the IP of the target system */ if ((ip = iface_ip2str(loaderData->netDev, AF_INET)) == NULL) { -- 1.6.0.3

1 1

[PATCH] include raid modules in mk-images and loader
by Joel Granados Moreno 16 Dec '08

16 Dec '08

Hi all: This is just lookin for a final sanity check of the two patches that include the dm modules in installer. there is still a standing issue with the first boot after installation but I'm pretty sure that its mkinitrelated, or dmraid related. In any case these modules best be in there so we can root out issues with the image growing a little. regards

2 4

[PATCH 0/4] Adding Support for EXTLINUX into grubby
by J.H. 16 Dec '08

16 Dec '08

This patch series add's support for EXTLINUX into grubby, adding --extlinux and --extlinuxmenu This is (1) a POC and initial implimentation, and (2) will be slightly updated to reflect some new changes that are going into SYSLINUX as I write this to simplify some of this code base. Also as a note the git tree is available at git://git.kernel.org/pub/scm/boot/warthog9/mkinitrd-syslinux.git - note this is only the branch that I've worked on, not a full tree you may want to add it in as a branch when you pull/clone. - John 'Warthog9' Hawley John 'Warthog9' Hawley (4): Add function definitions for addLine and addLineTmpl Fix not-found default entry when reading config Initial commit of EXTLINUX (no menu) support Initial commit of EXTLINUX (with menu) support grubby/grubby.c | 287 ++++++++++++++++++++++++++++++++++++++++++++++++++++-- 1 files changed, 276 insertions(+), 11 deletions(-)

2 5

[PATCH] Let X tell us when it's launched rather than just sleeping.
by Adam Jackson 16 Dec '08

16 Dec '08

X will send us SIGUSR1 when it's ready to accept connections. --- anaconda | 30 ++++++++++++++++++++++++++---- 1 files changed, 26 insertions(+), 4 deletions(-) diff --git a/anaconda b/anaconda index 2731c43..86d2205 100755 --- a/anaconda +++ b/anaconda @@ -832,11 +832,33 @@ if __name__ == "__main__": # running on a redirected X display, so start local X server if opts.display_mode == 'g' and not os.environ.has_key('DISPLAY') and not flags.usevnc: try: + # start X with its USR1 handler set to ignore. this will make it send + # us SIGUSR1 if it succeeds. if it fails, catch SIGCHLD and bomb out. + + def sigchld_handler(num, frame): + raise OSError + + def sigusr1_handler(num, frame): + pass + + def preexec_fn(): + signal.signal(signal.SIGUSR1, signal.SIG_IGN) + xout = open("/dev/tty5", "w") - proc = subprocess.Popen(["Xorg", "-br", "-logfile", "/tmp/X.log", ":1", "vt6", "-s", "1440", "-ac", "-nolisten", "tcp", "-dpi", "96"], close_fds=True, stdout=xout, stderr=xout) - time.sleep(5) - if proc.poll() is not None: - raise OSError + old_sigusr1 = signal.signal(signal.SIGUSR1, sigusr1_handler) + old_sigchld = signal.signal(signal.SIGCHLD, sigchld_handler) + + proc = subprocess.Popen(["Xorg", "-br", "-logfile", "/tmp/X.log", + ":1", "vt6", "-s", "1440", "-ac", + "-nolisten", "tcp", "-dpi", "96"], + close_fds=True, stdout=xout, stderr=xout, + preexec_fn=preexec_fn) + + signal.pause() + + signal.signal(signal.SIGUSR1, old_sigusr1) + signal.signal(signal.SIGCHLD, old_sigchld) + os.environ["DISPLAY"] = ":1" doStartupX11Actions(opts.runres) xserver_pid = proc.pid -- 1.6.0.4

2 2

[PATCH] Add --root-device to upgrade in ks.cfg
by Alexander Todorov 16 Dec '08

16 Dec '08

Hi all, the attached patches against anaconda and pykickstart add new option to the upgrade command in kickstart. upgrade [--root-device=/dev/sda2] The purpose is to be able to automatically upgrade systems that are dual boot. Specifying which is the root device makes upgrades more flexible. The patches however do not resolve a bug (or a feature) where if you have upgrade in your ks.cfg and do not specify the root device anaconda will mount the first in the list: 363 # XXX handle going backwards 364 def upgradeMountFilesystems(anaconda): 365 # mount everything and turn on swap 366 367 if flags.setupFilesystems: 368 try: 369 mountRootPartition(anaconda, anaconda.id.upgradeRoot[0], anaconda.id.fsset, 370 allowDirty = 0) Please review the patches and send me your feedback. Thanks, Alexander.

2 4

Re: [PATCH] Write /etc/resolv.conf and /etc/hosts in stage1 on s390.
by Joel Granados 16 Dec '08

16 Dec '08

The patch does not conflict, it fits comfortably on top of what dcantrell commited. What I see in the code is that dcantrell makes sure that the DNS var is Ok. And then we go an put those strings into /etc/hosts. To me it looks ok. The only strange thing I saw was that the $DNS var before this patch was in between '""' and they are taken away for some reason. Don't think that this will be a problem since a dns address with a space inside is just wrong. But if it bothers people I can put it back in :) FYI, I'm looking at git diff 1263e881d7029c85bb0cca42abd5851827751d17, from the current tree. Regards. ----- "Hans de Goede" <hdegoede(a)redhat.com> wrote: > David Cantrell wrote: > > Hans de Goede wrote: > >> > >> Joel Granados Moreno wrote: > >>> From: David Cantrell <dcantrell(a)redhat.com> > >>> > >>> Write out nameserver lines in /etc/resolv.conf correctly and > >>> write out /etc/hosts containing localhost lines as well as a > >>> line for the host itself. This allows ssh X forwarding to work > >>> and telnetd to work.(459730) > >>> --- > >>> loader2/linuxrc.s390 | 9 ++++++++- > >>> 1 files changed, 8 insertions(+), 1 deletions(-) > >> Hmm, this seems to clash with the patch from dcantrell I've just > >> reviewed, please coordinate. > > > > The patch you reviewed for me was for rhel4-branch only. I should > have > > specified that. > > > > So is Joel's patch :) > > Regards, > > Hans > > _______________________________________________ > Anaconda-devel-list mailing list > Anaconda-devel-list(a)redhat.com > https://www.redhat.com/mailman/listinfo/anaconda-devel-list -- Joel Andres Granados Red Hat / Brno Czech Republic

2 1

[PATCH] make sure that /etc/hosts is there at install time.
by Joel Granados Moreno 16 Dec '08

16 Dec '08

Hello list. This is for RHEL4. For x forwarding to work properly, the installer must have an /etc/host. This patch makes sure that this file gets created at init time. This patch was backported from RHEL5s 7b3bd255f424d3accee14a9bc11db7104028fa5e. Review greatly appreciated.

4 5

[PATCH] we should now load the newly included raid modules
by Joel Granados Moreno 16 Dec '08

16 Dec '08

This is a follow up to my previous addition to the initrd image. I added some raid modules, but never initialized them at init time. This patch corrects that behavior. Review appreciated

2 2

[PATCH] Allow empty DNS variable in s390 CMS conf file (#465175)
by David Cantrell 15 Dec '08

15 Dec '08

If the DNS variable contains an empty space or a colon, anaconda gives a traceback. --- loader2/linuxrc.s390 | 42 ++++++++++++++++++++++++++++++++++++++---- 1 files changed, 38 insertions(+), 4 deletions(-) diff --git a/loader2/linuxrc.s390 b/loader2/linuxrc.s390 index 0400ede..7b8fcde 100644 --- a/loader2/linuxrc.s390 +++ b/loader2/linuxrc.s390 @@ -60,6 +60,24 @@ checkipv4() return $? } +checkdns() +{ + dns=$1 + # DNS syntax is addr1:addr2:addr3 so first split by ":" + dnslist=$(echo $dns | awk -F ':' '{i=1; while (i<=NF) {print $i; i=i+1}}') + found=0 + for i in $dnslist; do + # note: only support ipv4 DNS currently, + # since both ipv6 address and DNS= parameter can contain ":" + checkipv4 $i + [ $? -eq 1 ] && return 1 + # track whether we've found at least one valid IP + found=1 + done + [ $found -eq 1 ] && return 0 + return 1 +} + doshutdown() { exec /sbin/shutdown @@ -457,13 +475,29 @@ route add default gw $GATEWAY dev $DEVICE 2>/dev/null # BH FIXME: Workaround for manual MACADDR, need ping to update arp table ping -c 1 $GATEWAY > /dev/null -if [ -z "$DNS" ]; then - echo $"Enter your DNS server(s), separated by colons (:):" +# DNS is optional, but if specified it must be correct +if [ -n "$DNS" ]; then + checkdns $DNS + ret=$? +fi +if [ -z "$DNS" ] || [ $ret -ne 0 ]; then + echo $"Enter your DNS server(s), separated by colons (:)" + echo $"Leave this blank and press enter to skip" read DNS + checkdns $DNS + ret=$? + while [ -n "$DNS" ] && [ $ret -ne 0 ]; do + echo -n "Invalid DNS address format. " + echo $"Enter your DNS server(s), separated by colons (:)" + echo $"Leave this blank and press enter to skip" + read DNS + checkdns $DNS + ret=$? + done fi if [ -z "$DNS" ]; then - echo $"You might encounter problems without a nameserver, especially" - echo $"with FTP installs" + echo $"WARNING: You might encounter problems without a nameserver," + echo $"especially with FTP installs" fi if [ -n "$DNS" -a -z "$SEARCHDNS" ]; then -- 1.6.0.3

2 1

Possible failure when checking "Updates" repository
by Jeroen van Meeuwen 15 Dec '08

15 Dec '08

Hi, Now that in anaconda, during package selection, you can enable the "Updates" repository for a given release, has anyone thought of the following scenario? - You select/enable the "Updates" repository - Included in your selection of packages is package foo, available in both the "Fedora/" repository as well as the "Updates" repository - Package foo (from "Updates") now requires package bar, which is in "Everything/" (not in "Fedora/"), but hasn't been updated since GA (e.g, is not available in the "Updates" repository) This would cause failed dependencies if the "Updates" repository is enabled during installation, also enable the "Everything" repository. Let me know if this doesn't make any sense. Kind regards, Jeroen van Meeuwen -kanarip

4 8

← Newer
1
2
3
4
5
6
7
8
9
10
Older →

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Anaconda-devel December 2008