10:14 a.m.
Download from: http://people.redhat.com/anderson
Changelog:
- Added "bison" to the BuildRequires line of the crash.spec file.
Without the patch, the build of the embedded gdb-7.6 module will fail
unless either /usr/bin/bison or /usr/bin/yacc are available. The
failure will result in a stream of error messages from different
files that indicate:
multiple definition of 'main'
undefined reference to 'c_parse_escape'
undefined reference to 'ada_parse'
undefined reference to 'ada_error'
undefined reference to 'c_parse'
undefined reference to 'c_error'
undefined reference to 'cp_demangled_name_to_comp'
undefined reference to 'cp_demangled_name_parse_free'
undefined reference to 'cp_comp_to_string'
undefined reference to 'cp_new_demangle_parse_info'
and the build fails like so:
collect2: ld returned 1 exit status
make[4]: *** [gdb] Error 1
crash build failed
If building with rpmbuild, the new BuildRequires "bison" entry will
prevent the build from initiating unless the bison package has been
installed. If building with the tar.gz file, the build attempt will
proceed and fail unless either the bison or byacc (Berkeley Yacc)
package is installed.
(anderson(a)redhat.com)
- Fix the S390X initialization sequence on kernels that are configured
with CONFIG_STRICT_DEVMEM to automatically try /proc/kcore if:
(1) the /dev/crash driver is not available, and
(2) the initial /dev/mem access fails.
Without the patch, if /dev/mem is selected as the memory source and
it is restricted, the crash session will fail during initialization
with the error message "crash: read error: kernel virtual address:
<address> type: cpu_possible_mask".
(anderson(a)redhat.com)
- When checking whether a argument on the crash command line is a
dumpfile that may be in makedumpfile's "flattened" format, do not
bother checking character device files.
(anderson(a)redhat.com)
- Fix for the PPC64 virtual-to-physical virtual address translation
mechanism for vmalloc and user-space virtual addresses on Linux 3.10
and later kernels. Without the patch, the message "WARNING: cannot
access vmalloc'd module memory" is displayed during initialization,
and during the crash session, if a command attempts to translate or
read a vmalloc or user-space virtual address, it will fail.
(anderson(a)redhat.com)
- Clean up all files that emit "warning: format not a string literal
and no format arguments" when compiled with -Wformat-security warning
option. All instances of fprintf, sprintf and snprintf using the
format "fprintf(fp, buf)" are replaced with "fprintf(fp, "%s",
buf)".
Also, the -Wformat-security warning option has been added to the
option list used when compiling with "make warn".
(stefan.bader(a)canonical.com, anderson(a)redhat.com)
- Fix a build failure when compiling with very old gcc-3.4.6 version
on a 2.6.9-based RHEL4 IA64 host. The bfd library in gdb-7.6 is
compiled with the -Werror option, and it fails with the message
"elflink.c:4733: warning: 'idx' might be used uninitialized in this
function".
(anderson(a)redhat.com)
- Fix a build failure when compiling with very old gcc-3.4.6 version
on a 2.6.9-based RHEL4 S390 or S390X hosts. The embedded gdb-7.6
fails to compile with the error message "s390-nat.c:364: error:
storage size of 'iov' isn't known".
(anderson(a)redhat.com)
- Fix to properly store two-digit kernel version numbers.
(timo.lindfors(a)iki.fi)
- Fix to provide hugepage address translation for the "vtop" command on
the PPC64 architecture.
(hbathini(a)linux.vnet.ibm.com)
- Fix for the "log" command to account for the kernel data structure
name change from "log" to "printk_log" in Linux 3.11-rc4 and later
kernels. Without the patch, the message "WARNING: log buf data
structure(s) have changed" will be displayed during initialization
and by the "log" command.
(anatol.pomozov(a)gmail.com)
- Fix to add a linefeed after the description of the "kmem -I" option
in the "help kmem" output, which was recently added in crash-7.0.0.
(anderson(a)redhat.com)
- Document the "-s" command line option in the "crash -h|--help"
output
and in the crash.8 man page to also indicate that runtime command
scrolling is turned off by default.
(anderson(a)redhat.com)
- Fix for the "irq -d" option on 2.6.25 and later X86_64 kernels to
display the Intel interrupt descriptor table contents. Without the
patch, those kernel versions would display "irq: -d option not
supported or applicable on this architecture or kernel".
(anatol.pomozov(a)gmail.com)
- Fix for the "kmem -[sS]" options on Linux 3.11-rc1 and later kernels
that are configured with CONFIG_SLAB. Without the patch, the command
fails with the error message "kmem: invalid structure member offset:
kmem_cache_s_lists".
(anatol.pomozov(a)gmail.com)
- Fix for the "kmem <address>" and the "bt -F" options on Linux
3.8
and later kernels that are configured with CONFIG_SLUB. Without the
patch, the command would fail with the error message "kmem: invalid
structure member offset: page_slab".
(anderson(a)redhat.com)
- Fix misspellings in the "bt" and "search" help page output.
(anatol.pomozov(a)gmail.com)
- Fix for the determination of the base of the kernel's unity-mapped
virtual address region on recent ARM kernels whose "_stext" variable
address has changed from 0xc0008000 to 0xc0100000. Without the
patch, the crash session fails during invocation with the error
message "crash: vmlinux and vmcore do not match!".
(Jan.Karlsson(a)sonymobile.com)
- When printing data structures, prevent the embedded gdb from
symbolically translating pointers that are not kernel virtual
addresses. Kernel or module symbols that are not virtual addresses
can be mistaken for virtual addresses, leading to NULL pointers
being invalidly translated into a symbol name from the vmlinux or
module object file. For example, in X86_64 kernels, NULL pointers
are translated into the symbol "irq_stack_union", whose value is
not a virtual address, but rather a per-cpu offset value of 0.
(anderson(a)redhat.com)
- Fix for the "kmem -s <address>" or "kmem <address>"
options on
Linux 3.11 and later kernels configured with CONFIG_SLAB. Without
the patch, both commands fail with the error message "kmem: cannot
resolve cache_cache".
(anderson(a)redhat.com)
- Fix to prevent the "bt" command from generating a segmentation
violation in a case where the per-cpu "current_task" variable and
the runqueue's "curr" variable did not agree, and the panic task
had overflowed its kernel stack. This led to the selection of the
a starting RSP address which belonged to the other task; without
the patch, the command generated a segmentation violation after
printing the first frame of the backtrace.
(anderson(a)redhat.com)
1:57 p.m.
New subject: crash-7.0.2 is available
Dave - I just fixed the warnings for eppic's compile.
http://code.google.com/p/eppic/source/detail?r=09e8b97c2743cedeafe5b17b4b...
________________________________________
From: crash-utility-bounces(a)redhat.com [crash-utility-bounces(a)redhat.com] on behalf of
Dave Anderson [anderson(a)redhat.com]
Sent: Wednesday, September 04, 2013 11:14 AM
To: Discussion list for crash utility usage, maintenance and development
Subject: [Crash-utility] [ANNOUNCE] crash-7.0.2 is available
Download from: http://people.redhat.com/anderson
Changelog:
- Added "bison" to the BuildRequires line of the crash.spec file.
Without the patch, the build of the embedded gdb-7.6 module will fail
unless either /usr/bin/bison or /usr/bin/yacc are available. The
failure will result in a stream of error messages from different
files that indicate:
multiple definition of 'main'
undefined reference to 'c_parse_escape'
undefined reference to 'ada_parse'
undefined reference to 'ada_error'
undefined reference to 'c_parse'
undefined reference to 'c_error'
undefined reference to 'cp_demangled_name_to_comp'
undefined reference to 'cp_demangled_name_parse_free'
undefined reference to 'cp_comp_to_string'
undefined reference to 'cp_new_demangle_parse_info'
and the build fails like so:
collect2: ld returned 1 exit status
make[4]: *** [gdb] Error 1
crash build failed
If building with rpmbuild, the new BuildRequires "bison" entry will
prevent the build from initiating unless the bison package has been
installed. If building with the tar.gz file, the build attempt will
proceed and fail unless either the bison or byacc (Berkeley Yacc)
package is installed.
(anderson(a)redhat.com)
- Fix the S390X initialization sequence on kernels that are configured
with CONFIG_STRICT_DEVMEM to automatically try /proc/kcore if:
(1) the /dev/crash driver is not available, and
(2) the initial /dev/mem access fails.
Without the patch, if /dev/mem is selected as the memory source and
it is restricted, the crash session will fail during initialization
with the error message "crash: read error: kernel virtual address:
<address> type: cpu_possible_mask".
(anderson(a)redhat.com)
- When checking whether a argument on the crash command line is a
dumpfile that may be in makedumpfile's "flattened" format, do not
bother checking character device files.
(anderson(a)redhat.com)
- Fix for the PPC64 virtual-to-physical virtual address translation
mechanism for vmalloc and user-space virtual addresses on Linux 3.10
and later kernels. Without the patch, the message "WARNING: cannot
access vmalloc'd module memory" is displayed during initialization,
and during the crash session, if a command attempts to translate or
read a vmalloc or user-space virtual address, it will fail.
(anderson(a)redhat.com)
- Clean up all files that emit "warning: format not a string literal
and no format arguments" when compiled with -Wformat-security warning
option. All instances of fprintf, sprintf and snprintf using the
format "fprintf(fp, buf)" are replaced with "fprintf(fp, "%s",
buf)".
Also, the -Wformat-security warning option has been added to the
option list used when compiling with "make warn".
(stefan.bader(a)canonical.com, anderson(a)redhat.com)
- Fix a build failure when compiling with very old gcc-3.4.6 version
on a 2.6.9-based RHEL4 IA64 host. The bfd library in gdb-7.6 is
compiled with the -Werror option, and it fails with the message
"elflink.c:4733: warning: 'idx' might be used uninitialized in this
function".
(anderson(a)redhat.com)
- Fix a build failure when compiling with very old gcc-3.4.6 version
on a 2.6.9-based RHEL4 S390 or S390X hosts. The embedded gdb-7.6
fails to compile with the error message "s390-nat.c:364: error:
storage size of 'iov' isn't known".
(anderson(a)redhat.com)
- Fix to properly store two-digit kernel version numbers.
(timo.lindfors(a)iki.fi)
- Fix to provide hugepage address translation for the "vtop" command on
the PPC64 architecture.
(hbathini(a)linux.vnet.ibm.com)
- Fix for the "log" command to account for the kernel data structure
name change from "log" to "printk_log" in Linux 3.11-rc4 and later
kernels. Without the patch, the message "WARNING: log buf data
structure(s) have changed" will be displayed during initialization
and by the "log" command.
(anatol.pomozov(a)gmail.com)
- Fix to add a linefeed after the description of the "kmem -I" option
in the "help kmem" output, which was recently added in crash-7.0.0.
(anderson(a)redhat.com)
- Document the "-s" command line option in the "crash -h|--help"
output
and in the crash.8 man page to also indicate that runtime command
scrolling is turned off by default.
(anderson(a)redhat.com)
- Fix for the "irq -d" option on 2.6.25 and later X86_64 kernels to
display the Intel interrupt descriptor table contents. Without the
patch, those kernel versions would display "irq: -d option not
supported or applicable on this architecture or kernel".
(anatol.pomozov(a)gmail.com)
- Fix for the "kmem -[sS]" options on Linux 3.11-rc1 and later kernels
that are configured with CONFIG_SLAB. Without the patch, the command
fails with the error message "kmem: invalid structure member offset:
kmem_cache_s_lists".
(anatol.pomozov(a)gmail.com)
- Fix for the "kmem <address>" and the "bt -F" options on Linux
3.8
and later kernels that are configured with CONFIG_SLUB. Without the
patch, the command would fail with the error message "kmem: invalid
structure member offset: page_slab".
(anderson(a)redhat.com)
- Fix misspellings in the "bt" and "search" help page output.
(anatol.pomozov(a)gmail.com)
- Fix for the determination of the base of the kernel's unity-mapped
virtual address region on recent ARM kernels whose "_stext" variable
address has changed from 0xc0008000 to 0xc0100000. Without the
patch, the crash session fails during invocation with the error
message "crash: vmlinux and vmcore do not match!".
(Jan.Karlsson(a)sonymobile.com)
- When printing data structures, prevent the embedded gdb from
symbolically translating pointers that are not kernel virtual
addresses. Kernel or module symbols that are not virtual addresses
can be mistaken for virtual addresses, leading to NULL pointers
being invalidly translated into a symbol name from the vmlinux or
module object file. For example, in X86_64 kernels, NULL pointers
are translated into the symbol "irq_stack_union", whose value is
not a virtual address, but rather a per-cpu offset value of 0.
(anderson(a)redhat.com)
- Fix for the "kmem -s <address>" or "kmem <address>"
options on
Linux 3.11 and later kernels configured with CONFIG_SLAB. Without
the patch, both commands fail with the error message "kmem: cannot
resolve cache_cache".
(anderson(a)redhat.com)
- Fix to prevent the "bt" command from generating a segmentation
violation in a case where the per-cpu "current_task" variable and
the runqueue's "curr" variable did not agree, and the panic task
had overflowed its kernel stack. This led to the selection of the
a starting RSP address which belonged to the other task; without
the patch, the command generated a segmentation violation after
printing the first frame of the backtrace.
(anderson(a)redhat.com)
--
Crash-utility mailing list
Crash-utility(a)redhat.com
https://www.redhat.com/mailman/listinfo/crash-utility
2:35 p.m.
New subject: crash-7.0.2 is available
----- Original Message -----
Dave - I just fixed the warnings for eppic's compile.
http://code.google.com/p/eppic/source/detail?r=09e8b97c2743cedeafe5b17b4b...
Builds clean as a whistle -- thanks!
Dave
________________________________________
From: crash-utility-bounces(a)redhat.com [crash-utility-bounces(a)redhat.com] on
behalf of Dave Anderson [anderson(a)redhat.com]
Sent: Wednesday, September 04, 2013 11:14 AM
To: Discussion list for crash utility usage, maintenance and development
Subject: [Crash-utility] [ANNOUNCE] crash-7.0.2 is available
Download from: http://people.redhat.com/anderson
Changelog:
- Added "bison" to the BuildRequires line of the crash.spec file.
Without the patch, the build of the embedded gdb-7.6 module will fail
unless either /usr/bin/bison or /usr/bin/yacc are available. The
failure will result in a stream of error messages from different
files that indicate:
multiple definition of 'main'
undefined reference to 'c_parse_escape'
undefined reference to 'ada_parse'
undefined reference to 'ada_error'
undefined reference to 'c_parse'
undefined reference to 'c_error'
undefined reference to 'cp_demangled_name_to_comp'
undefined reference to 'cp_demangled_name_parse_free'
undefined reference to 'cp_comp_to_string'
undefined reference to 'cp_new_demangle_parse_info'
and the build fails like so:
collect2: ld returned 1 exit status
make[4]: *** [gdb] Error 1
crash build failed
If building with rpmbuild, the new BuildRequires "bison" entry will
prevent the build from initiating unless the bison package has been
installed. If building with the tar.gz file, the build attempt will
proceed and fail unless either the bison or byacc (Berkeley Yacc)
package is installed.
(anderson(a)redhat.com)
- Fix the S390X initialization sequence on kernels that are configured
with CONFIG_STRICT_DEVMEM to automatically try /proc/kcore if:
(1) the /dev/crash driver is not available, and
(2) the initial /dev/mem access fails.
Without the patch, if /dev/mem is selected as the memory source and
it is restricted, the crash session will fail during initialization
with the error message "crash: read error: kernel virtual address:
<address> type: cpu_possible_mask".
(anderson(a)redhat.com)
- When checking whether a argument on the crash command line is a
dumpfile that may be in makedumpfile's "flattened" format, do not
bother checking character device files.
(anderson(a)redhat.com)
- Fix for the PPC64 virtual-to-physical virtual address translation
mechanism for vmalloc and user-space virtual addresses on Linux 3.10
and later kernels. Without the patch, the message "WARNING: cannot
access vmalloc'd module memory" is displayed during initialization,
and during the crash session, if a command attempts to translate or
read a vmalloc or user-space virtual address, it will fail.
(anderson(a)redhat.com)
- Clean up all files that emit "warning: format not a string literal
and no format arguments" when compiled with -Wformat-security warning
option. All instances of fprintf, sprintf and snprintf using the
format "fprintf(fp, buf)" are replaced with "fprintf(fp,
"%s", buf)".
Also, the -Wformat-security warning option has been added to the
option list used when compiling with "make warn".
(stefan.bader(a)canonical.com, anderson(a)redhat.com)
- Fix a build failure when compiling with very old gcc-3.4.6 version
on a 2.6.9-based RHEL4 IA64 host. The bfd library in gdb-7.6 is
compiled with the -Werror option, and it fails with the message
"elflink.c:4733: warning: 'idx' might be used uninitialized in this
function".
(anderson(a)redhat.com)
- Fix a build failure when compiling with very old gcc-3.4.6 version
on a 2.6.9-based RHEL4 S390 or S390X hosts. The embedded gdb-7.6
fails to compile with the error message "s390-nat.c:364: error:
storage size of 'iov' isn't known".
(anderson(a)redhat.com)
- Fix to properly store two-digit kernel version numbers.
(timo.lindfors(a)iki.fi)
- Fix to provide hugepage address translation for the "vtop" command on
the PPC64 architecture.
(hbathini(a)linux.vnet.ibm.com)
- Fix for the "log" command to account for the kernel data structure
name change from "log" to "printk_log" in Linux 3.11-rc4 and
later
kernels. Without the patch, the message "WARNING: log buf data
structure(s) have changed" will be displayed during initialization
and by the "log" command.
(anatol.pomozov(a)gmail.com)
- Fix to add a linefeed after the description of the "kmem -I" option
in the "help kmem" output, which was recently added in crash-7.0.0.
(anderson(a)redhat.com)
- Document the "-s" command line option in the "crash -h|--help"
output
and in the crash.8 man page to also indicate that runtime command
scrolling is turned off by default.
(anderson(a)redhat.com)
- Fix for the "irq -d" option on 2.6.25 and later X86_64 kernels to
display the Intel interrupt descriptor table contents. Without the
patch, those kernel versions would display "irq: -d option not
supported or applicable on this architecture or kernel".
(anatol.pomozov(a)gmail.com)
- Fix for the "kmem -[sS]" options on Linux 3.11-rc1 and later kernels
that are configured with CONFIG_SLAB. Without the patch, the command
fails with the error message "kmem: invalid structure member offset:
kmem_cache_s_lists".
(anatol.pomozov(a)gmail.com)
- Fix for the "kmem <address>" and the "bt -F" options on
Linux 3.8
and later kernels that are configured with CONFIG_SLUB. Without the
patch, the command would fail with the error message "kmem: invalid
structure member offset: page_slab".
(anderson(a)redhat.com)
- Fix misspellings in the "bt" and "search" help page output.
(anatol.pomozov(a)gmail.com)
- Fix for the determination of the base of the kernel's unity-mapped
virtual address region on recent ARM kernels whose "_stext" variable
address has changed from 0xc0008000 to 0xc0100000. Without the
patch, the crash session fails during invocation with the error
message "crash: vmlinux and vmcore do not match!".
(Jan.Karlsson(a)sonymobile.com)
- When printing data structures, prevent the embedded gdb from
symbolically translating pointers that are not kernel virtual
addresses. Kernel or module symbols that are not virtual addresses
can be mistaken for virtual addresses, leading to NULL pointers
being invalidly translated into a symbol name from the vmlinux or
module object file. For example, in X86_64 kernels, NULL pointers
are translated into the symbol "irq_stack_union", whose value is
not a virtual address, but rather a per-cpu offset value of 0.
(anderson(a)redhat.com)
- Fix for the "kmem -s <address>" or "kmem <address>"
options on
Linux 3.11 and later kernels configured with CONFIG_SLAB. Without
the patch, both commands fail with the error message "kmem: cannot
resolve cache_cache".
(anderson(a)redhat.com)
- Fix to prevent the "bt" command from generating a segmentation
violation in a case where the per-cpu "current_task" variable and
the runqueue's "curr" variable did not agree, and the panic task
had overflowed its kernel stack. This led to the selection of the
a starting RSP address which belonged to the other task; without
the patch, the command generated a segmentation violation after
printing the first frame of the backtrace.
(anderson(a)redhat.com)
--
Crash-utility mailing list
Crash-utility(a)redhat.com
https://www.redhat.com/mailman/listinfo/crash-utility
--
Crash-utility mailing list
Crash-utility(a)redhat.com
https://www.redhat.com/mailman/listinfo/crash-utility
4444
days inactive
4444
days old
devel@lists.crash-utility.osci.io
2 comments
2 participants
participants (2)
-
Dave Anderson -
Luc Chouinard