8:47 a.m.
Qemu migration code added new sections to add features
for live migration of VM. For loading vmcore file captured
with 'virsh dump' we need to parse these sections in crash.
This series contains two patches, which parse these sections:
patch1: parse 'vm_configuration' section
patch2: parse 'vm_footer' section
qemu-load.c | 20 ++++++++++++++++++++
qemu-load.h | 4 +++-
2 files changed, 23 insertions(+), 1 deletion(-)
8:47 a.m.
New subject: [PATCH 1/2] qemu: parse 'vm_configuration' section
Teach crash to parse 'QEMU_VM_CONFIGURATION' section. This
section is added by migration code to save VM running state.
Signed-off-by: Pankaj Gupta <pagupta(a)redhat.com>
---
qemu-load.c | 16 ++++++++++++++++
qemu-load.h | 3 ++-
2 files changed, 18 insertions(+), 1 deletion(-)
diff --git a/qemu-load.c b/qemu-load.c
index 8b4372e..7ee58a3 100644
--- a/qemu-load.c
+++ b/qemu-load.c
@@ -208,6 +208,16 @@ get_string (FILE *fp, char *name)
name[sz] = 0;
return sz;
}
+static int
+get_string_len (FILE *fp, char *name, uint32_t sz)
+{
+ size_t items ATTRIBUTE_UNUSED;
+ if (sz == EOF)
+ return -1;
+ items = fread (name, sz, 1, fp);
+ name[sz] = 0;
+ return sz;
+}
static void
ram_read_blocks (FILE *fp, uint64_t size)
@@ -924,6 +934,7 @@ qemu_load (const struct qemu_device_loader *devices, uint32_t
required_features,
struct qemu_device_list *result = NULL;
struct qemu_device *last = NULL;;
size_t items ATTRIBUTE_UNUSED;
+ char name[257];
switch (get_be32 (fp)) {
case QEMU_VM_FILE_MAGIC:
@@ -961,6 +972,11 @@ qemu_load (const struct qemu_device_loader *devices, uint32_t
required_features,
break;
if (sec == QEMU_VM_EOF)
break;
+ if (sec == QEMU_VM_CONFIGURATION) {
+ uint32_t len = get_be32 (fp);
+ get_string_len (fp, name, len);
+ continue;
+ }
d = device_get (devices, result, sec, fp);
if (!d)
diff --git a/qemu-load.h b/qemu-load.h
index e52b068..d5b2078 100644
--- a/qemu-load.h
+++ b/qemu-load.h
@@ -29,7 +29,8 @@ enum qemu_save_section {
QEMU_VM_SECTION_PART,
QEMU_VM_SECTION_END,
QEMU_VM_SECTION_FULL,
- QEMU_VM_SUBSECTION
+ QEMU_VM_SUBSECTION,
+ QEMU_VM_CONFIGURATION = 0x07
};
enum qemu_features {
--
1.9.3
5:25 a.m.
New subject: [PATCH 1/2] qemu: parse 'vm_configuration' section
* Pankaj Gupta (pagupta(a)redhat.com) wrote:
Teach crash to parse 'QEMU_VM_CONFIGURATION' section. This
section is added by migration code to save VM running state.
Signed-off-by: Pankaj Gupta <pagupta(a)redhat.com>
---
qemu-load.c | 16 ++++++++++++++++
qemu-load.h | 3 ++-
2 files changed, 18 insertions(+), 1 deletion(-)
diff --git a/qemu-load.c b/qemu-load.c
index 8b4372e..7ee58a3 100644
--- a/qemu-load.c
+++ b/qemu-load.c
@@ -208,6 +208,16 @@ get_string (FILE *fp, char *name)
name[sz] = 0;
return sz;
}
+static int
+get_string_len (FILE *fp, char *name, uint32_t sz)
+{
+ size_t items ATTRIBUTE_UNUSED;
+ if (sz == EOF)
+ return -1;
+ items = fread (name, sz, 1, fp);
+ name[sz] = 0;
+ return sz;
+}
static void
ram_read_blocks (FILE *fp, uint64_t size)
@@ -924,6 +934,7 @@ qemu_load (const struct qemu_device_loader *devices, uint32_t
required_features,
struct qemu_device_list *result = NULL;
struct qemu_device *last = NULL;;
size_t items ATTRIBUTE_UNUSED;
+ char name[257];
switch (get_be32 (fp)) {
case QEMU_VM_FILE_MAGIC:
@@ -961,6 +972,11 @@ qemu_load (const struct qemu_device_loader *devices, uint32_t
required_features,
break;
if (sec == QEMU_VM_EOF)
break;
+ if (sec == QEMU_VM_CONFIGURATION) {
+ uint32_t len = get_be32 (fp);
+ get_string_len (fp, name, len);
+ continue;
+ }
Be careful about buffer over runs! That 'len' must be checked to ensure
it fits into 'name' - or dynamically allocate 'name'. OK, in practice the
value
will be small, but this tool is aimed to read crash files that might have been
dumped by a customer and so you have to treat the contents of the file
as potentially hostile, either due to file corruption or really someone trying
to be nasty.
Dave
d = device_get (devices, result, sec, fp);
if (!d)
diff --git a/qemu-load.h b/qemu-load.h
index e52b068..d5b2078 100644
--- a/qemu-load.h
+++ b/qemu-load.h
@@ -29,7 +29,8 @@ enum qemu_save_section {
QEMU_VM_SECTION_PART,
QEMU_VM_SECTION_END,
QEMU_VM_SECTION_FULL,
- QEMU_VM_SUBSECTION
+ QEMU_VM_SUBSECTION,
+ QEMU_VM_CONFIGURATION = 0x07
};
enum qemu_features {
--
1.9.3
--
Dr. David Alan Gilbert / dgilbert(a)redhat.com / Manchester, UK
6:43 a.m.
New subject: [PATCH 1/2] qemu: parse 'vm_configuration' section
* Pankaj Gupta (pagupta(a)redhat.com) wrote:
> Teach crash to parse 'QEMU_VM_CONFIGURATION' section. This
> section is added by migration code to save VM running state.
>
> Signed-off-by: Pankaj Gupta <pagupta(a)redhat.com>
> ---
> qemu-load.c | 16 ++++++++++++++++
> qemu-load.h | 3 ++-
> 2 files changed, 18 insertions(+), 1 deletion(-)
>
> diff --git a/qemu-load.c b/qemu-load.c
> index 8b4372e..7ee58a3 100644
> --- a/qemu-load.c
> +++ b/qemu-load.c
> @@ -208,6 +208,16 @@ get_string (FILE *fp, char *name)
> name[sz] = 0;
> return sz;
> }
> +static int
> +get_string_len (FILE *fp, char *name, uint32_t sz)
> +{
> + size_t items ATTRIBUTE_UNUSED;
> + if (sz == EOF)
> + return -1;
> + items = fread (name, sz, 1, fp);
> + name[sz] = 0;
> + return sz;
> +}
>
> static void
> ram_read_blocks (FILE *fp, uint64_t size)
> @@ -924,6 +934,7 @@ qemu_load (const struct qemu_device_loader *devices,
> uint32_t required_features,
> struct qemu_device_list *result = NULL;
> struct qemu_device *last = NULL;;
> size_t items ATTRIBUTE_UNUSED;
> + char name[257];
>
> switch (get_be32 (fp)) {
> case QEMU_VM_FILE_MAGIC:
> @@ -961,6 +972,11 @@ qemu_load (const struct qemu_device_loader *devices,
> uint32_t required_features,
> break;
> if (sec == QEMU_VM_EOF)
> break;
> + if (sec == QEMU_VM_CONFIGURATION) {
> + uint32_t len = get_be32 (fp);
> + get_string_len (fp, name, len);
> + continue;
> + }
Be careful about buffer over runs! That 'len' must be checked to ensure
it fits into 'name' - or dynamically allocate 'name'. OK, in practice
the
value
will be small, but this tool is aimed to read crash files that might have
been
dumped by a customer and so you have to treat the contents of the file
as potentially hostile, either due to file corruption or really someone
trying
to be nasty.
I agree with you. But I still not sure why in Qemu we selected 'uint32_t' for
length and why its not consistent everywhere? If in good case we expect 'length >
256'
bytes I will do the change and spin new version.
And if we are sure length will be '<=256' bytes, I think validating
'footerSecId' in
a new patch should avoid us parsing corrupt file. What you suggest here?
Dave
> d = device_get (devices, result, sec, fp);
> if (!d)
> diff --git a/qemu-load.h b/qemu-load.h
> index e52b068..d5b2078 100644
> --- a/qemu-load.h
> +++ b/qemu-load.h
> @@ -29,7 +29,8 @@ enum qemu_save_section {
> QEMU_VM_SECTION_PART,
> QEMU_VM_SECTION_END,
> QEMU_VM_SECTION_FULL,
> - QEMU_VM_SUBSECTION
> + QEMU_VM_SUBSECTION,
> + QEMU_VM_CONFIGURATION = 0x07
> };
>
> enum qemu_features {
> --
> 1.9.3
>
--
Dr. David Alan Gilbert / dgilbert(a)redhat.com / Manchester, UK
6:57 a.m.
New subject: [PATCH 1/2] qemu: parse 'vm_configuration' section
* Pankaj Gupta (pagupta(a)redhat.com) wrote:
>
> * Pankaj Gupta (pagupta(a)redhat.com) wrote:
> > Teach crash to parse 'QEMU_VM_CONFIGURATION' section. This
> > section is added by migration code to save VM running state.
> >
> > Signed-off-by: Pankaj Gupta <pagupta(a)redhat.com>
> > ---
> > qemu-load.c | 16 ++++++++++++++++
> > qemu-load.h | 3 ++-
> > 2 files changed, 18 insertions(+), 1 deletion(-)
> >
> > diff --git a/qemu-load.c b/qemu-load.c
> > index 8b4372e..7ee58a3 100644
> > --- a/qemu-load.c
> > +++ b/qemu-load.c
> > @@ -208,6 +208,16 @@ get_string (FILE *fp, char *name)
> > name[sz] = 0;
> > return sz;
> > }
> > +static int
> > +get_string_len (FILE *fp, char *name, uint32_t sz)
> > +{
> > + size_t items ATTRIBUTE_UNUSED;
> > + if (sz == EOF)
> > + return -1;
> > + items = fread (name, sz, 1, fp);
> > + name[sz] = 0;
> > + return sz;
> > +}
> >
> > static void
> > ram_read_blocks (FILE *fp, uint64_t size)
> > @@ -924,6 +934,7 @@ qemu_load (const struct qemu_device_loader *devices,
> > uint32_t required_features,
> > struct qemu_device_list *result = NULL;
> > struct qemu_device *last = NULL;;
> > size_t items ATTRIBUTE_UNUSED;
> > + char name[257];
> >
> > switch (get_be32 (fp)) {
> > case QEMU_VM_FILE_MAGIC:
> > @@ -961,6 +972,11 @@ qemu_load (const struct qemu_device_loader *devices,
> > uint32_t required_features,
> > break;
> > if (sec == QEMU_VM_EOF)
> > break;
> > + if (sec == QEMU_VM_CONFIGURATION) {
> > + uint32_t len = get_be32 (fp);
> > + get_string_len (fp, name, len);
> > + continue;
> > + }
>
> Be careful about buffer over runs! That 'len' must be checked to ensure
> it fits into 'name' - or dynamically allocate 'name'. OK, in
practice the
> value
> will be small, but this tool is aimed to read crash files that might have
> been
> dumped by a customer and so you have to treat the contents of the file
> as potentially hostile, either due to file corruption or really someone
> trying
> to be nasty.
I agree with you. But I still not sure why in Qemu we selected 'uint32_t' for
length and why its not consistent everywhere? If in good case we expect 'length >
256'
bytes I will do the change and spin new version.
No I don't know why we used uint32_t for it in qemu; I suspect it was just the
easiest
thing in the code; but I'd expect it to always be less than 256 in practice.
And if we are sure length will be '<=256' bytes, I think
validating 'footerSecId' in
a new patch should avoid us parsing corrupt file. What you suggest here?
The section ID and the length are two separate issues.
I think you *need* to fix it to either check the length or dynamically allocate it.
Checking footerSecID is a separate idea and something you might try but I'm not sure
how easy it would be to fit into crash's code.
Dave
>
> Dave
>
> > d = device_get (devices, result, sec, fp);
> > if (!d)
> > diff --git a/qemu-load.h b/qemu-load.h
> > index e52b068..d5b2078 100644
> > --- a/qemu-load.h
> > +++ b/qemu-load.h
> > @@ -29,7 +29,8 @@ enum qemu_save_section {
> > QEMU_VM_SECTION_PART,
> > QEMU_VM_SECTION_END,
> > QEMU_VM_SECTION_FULL,
> > - QEMU_VM_SUBSECTION
> > + QEMU_VM_SUBSECTION,
> > + QEMU_VM_CONFIGURATION = 0x07
> > };
> >
> > enum qemu_features {
> > --
> > 1.9.3
> >
> --
> Dr. David Alan Gilbert / dgilbert(a)redhat.com / Manchester, UK
>
--
Dr. David Alan Gilbert / dgilbert(a)redhat.com / Manchester, UK
8:47 a.m.
New subject: [PATCH 2/2] qemu: parse 'vm_footer' section
Parse 'QEMU_VM_SECTION_FOOTER' in crash. This section is also
added by VM migration code to avoid misparsing.
Signed-off-by: Pankaj Gupta <pagupta(a)redhat.com>
---
qemu-load.c | 4 ++++
qemu-load.h | 3 ++-
2 files changed, 6 insertions(+), 1 deletion(-)
diff --git a/qemu-load.c b/qemu-load.c
index 7ee58a3..4b8313d 100644
--- a/qemu-load.c
+++ b/qemu-load.c
@@ -972,6 +972,10 @@ qemu_load (const struct qemu_device_loader *devices, uint32_t
required_features,
break;
if (sec == QEMU_VM_EOF)
break;
+ if (sec == QEMU_VM_SECTION_FOOTER) {
+ uint32_t footerSecId = get_be32 (fp);
+ continue;
+ }
if (sec == QEMU_VM_CONFIGURATION) {
uint32_t len = get_be32 (fp);
get_string_len (fp, name, len);
diff --git a/qemu-load.h b/qemu-load.h
index d5b2078..f4fbb06 100644
--- a/qemu-load.h
+++ b/qemu-load.h
@@ -30,7 +30,8 @@ enum qemu_save_section {
QEMU_VM_SECTION_END,
QEMU_VM_SECTION_FULL,
QEMU_VM_SUBSECTION,
- QEMU_VM_CONFIGURATION = 0x07
+ QEMU_VM_CONFIGURATION = 0x07,
+ QEMU_VM_SECTION_FOOTER = 0x7e
};
enum qemu_features {
--
1.9.3
5:33 a.m.
New subject: [PATCH 2/2] qemu: parse 'vm_footer' section
* Pankaj Gupta (pagupta(a)redhat.com) wrote:
Parse 'QEMU_VM_SECTION_FOOTER' in crash. This section is
also
added by VM migration code to avoid misparsing.
Signed-off-by: Pankaj Gupta <pagupta(a)redhat.com>
---
qemu-load.c | 4 ++++
qemu-load.h | 3 ++-
2 files changed, 6 insertions(+), 1 deletion(-)
diff --git a/qemu-load.c b/qemu-load.c
index 7ee58a3..4b8313d 100644
--- a/qemu-load.c
+++ b/qemu-load.c
@@ -972,6 +972,10 @@ qemu_load (const struct qemu_device_loader *devices, uint32_t
required_features,
break;
if (sec == QEMU_VM_EOF)
break;
+ if (sec == QEMU_VM_SECTION_FOOTER) {
+ uint32_t footerSecId = get_be32 (fp);
+ continue;
+ }
Yes, that should work.
One possible thing you could add in the future would be to compare the
footerSecId with the ID in the last received section - they should match, if
they don't then you could print a warning that you think you might have a corrupt
file.
Dave
if (sec == QEMU_VM_CONFIGURATION) {
uint32_t len = get_be32 (fp);
get_string_len (fp, name, len);
diff --git a/qemu-load.h b/qemu-load.h
index d5b2078..f4fbb06 100644
--- a/qemu-load.h
+++ b/qemu-load.h
@@ -30,7 +30,8 @@ enum qemu_save_section {
QEMU_VM_SECTION_END,
QEMU_VM_SECTION_FULL,
QEMU_VM_SUBSECTION,
- QEMU_VM_CONFIGURATION = 0x07
+ QEMU_VM_CONFIGURATION = 0x07,
+ QEMU_VM_SECTION_FOOTER = 0x7e
};
enum qemu_features {
--
1.9.3
--
Dr. David Alan Gilbert / dgilbert(a)redhat.com / Manchester, UK
9:07 a.m.
New subject: [PATCH 0/2] qemu: Parse necessary sections in crash added in qemu
----- Original Message -----
Qemu migration code added new sections to add features
for live migration of VM. For loading vmcore file captured
with 'virsh dump' we need to parse these sections in crash.
This series contains two patches, which parse these sections:
patch1: parse 'vm_configuration' section
patch2: parse 'vm_footer' section
qemu-load.c | 20 ++++++++++++++++++++
qemu-load.h | 4 +++-
2 files changed, 23 insertions(+), 1 deletion(-)
--
Crash-utility mailing list
Crash-utility(a)redhat.com
https://www.redhat.com/mailman/listinfo/crash-utility
Hi Pankaj,
I suppose I can take the patch. But, given the advent of "virsh dump
--memory-only"
back in 2012, and more recently, Oleg's added support for QEMU ramdumps by specifying
a memory-backend-file object to the qemu-kvm command:
commit 89ed9d0a7f7da4578294a492c1ad857244ce7352
Author: Dave Anderson <anderson(a)redhat.com>
Date: Wed May 4 11:50:19 2016 -0400
Introduction of support for "live" ramdump files, such as those that
are specified by the QEMU mem-path argument of a memory-backend-file
object. This allows the running of a live crash session against a
QEMU guest from the host machine. In this example, the /tmp/MEM file
on a QEMU host represents the guest's physical memory:
$ qemu-kvm ...other-options... \
-object memory-backend-file,id=MEM,size=128m,mem-path=/tmp/MEM,share=on \
-numa node,memdev=MEM -m 128
and a live session run can be run against the guest kernel like so:
$ crash <path-to-guest-vmlinux> live:/tmp/MEM@0
By prepending the ramdump image name with "live:", the crash session will
act as if it were running a normal live session.
(oleg(a)redhat.com)
And if the guest above crashed, the leftover file can be used as a regular
ramdump dumpfile:
commit 67a815b8749fbd5c99c29d24b4e699b1d618ddbf
Author: Dave Anderson <anderson(a)redhat.com>
Date: Wed May 4 14:34:53 2016 -0400
Added support for x86_64 ramdump files. Without the patch, the crash
session fails immediately with the message "ramdump: unsupported
machine type: X86_64".
(anderson(a)redhat.com)
So I don't know why anybody would ever want to use the old "kvmdump"
format?
It's for all practical purposes deprecated, not to mention that it's really
an unsuitable excuse for a crash dumpfile.
Dave
9:23 a.m.
New subject: [PATCH 0/2] qemu: Parse necessary sections in crash added in qemu
BTW, this footerSecId variable should be declared as ATTRIBUTE_UNUSED:
$ make warn
... [ cut ] ...
cc -c -g -DX86_64 -DLZO -DSNAPPY -DGDB_7_6 qemu-load.c -Wall -O2 -Wstrict-prototypes
-Wmissing-prototypes -fstack-protector -Wformat-security
qemu-load.c: In function 'qemu_load':
qemu-load.c:976:13: warning: unused variable 'footerSecId' [-Wunused-variable]
uint32_t footerSecId = get_be32 (fp);
...
Dave
----- Original Message -----
----- Original Message -----
> Qemu migration code added new sections to add features
> for live migration of VM. For loading vmcore file captured
> with 'virsh dump' we need to parse these sections in crash.
>
> This series contains two patches, which parse these sections:
>
> patch1: parse 'vm_configuration' section
> patch2: parse 'vm_footer' section
>
> qemu-load.c | 20 ++++++++++++++++++++
> qemu-load.h | 4 +++-
> 2 files changed, 23 insertions(+), 1 deletion(-)
>
> --
> Crash-utility mailing list
> Crash-utility(a)redhat.com
> https://www.redhat.com/mailman/listinfo/crash-utility
>
Hi Pankaj,
I suppose I can take the patch. But, given the advent of "virsh dump
--memory-only"
back in 2012, and more recently, Oleg's added support for QEMU ramdumps by
specifying
a memory-backend-file object to the qemu-kvm command:
commit 89ed9d0a7f7da4578294a492c1ad857244ce7352
Author: Dave Anderson <anderson(a)redhat.com>
Date: Wed May 4 11:50:19 2016 -0400
Introduction of support for "live" ramdump files, such as those that
are specified by the QEMU mem-path argument of a memory-backend-file
object. This allows the running of a live crash session against a
QEMU guest from the host machine. In this example, the /tmp/MEM file
on a QEMU host represents the guest's physical memory:
$ qemu-kvm ...other-options... \
-object memory-backend-file,id=MEM,size=128m,mem-path=/tmp/MEM,share=on
\
-numa node,memdev=MEM -m 128
and a live session run can be run against the guest kernel like so:
$ crash <path-to-guest-vmlinux> live:/tmp/MEM@0
By prepending the ramdump image name with "live:", the crash session will
act as if it were running a normal live session.
(oleg(a)redhat.com)
And if the guest above crashed, the leftover file can be used as a regular
ramdump dumpfile:
commit 67a815b8749fbd5c99c29d24b4e699b1d618ddbf
Author: Dave Anderson <anderson(a)redhat.com>
Date: Wed May 4 14:34:53 2016 -0400
Added support for x86_64 ramdump files. Without the patch, the crash
session fails immediately with the message "ramdump: unsupported
machine type: X86_64".
(anderson(a)redhat.com)
So I don't know why anybody would ever want to use the old "kvmdump"
format?
It's for all practical purposes deprecated, not to mention that it's really
an unsuitable excuse for a crash dumpfile.
Dave
--
Crash-utility mailing list
Crash-utility(a)redhat.com
https://www.redhat.com/mailman/listinfo/crash-utility
10:16 a.m.
New subject: [PATCH 0/2] qemu: Parse necessary sections in crash added in qemu
Hi Pankaj,
Queued for crash-7.1.6:
https://github.com/crash-utility/crash/commit/8cb8e85de65c4b51ca7537186e6...
Dave
----- Original Message -----
BTW, this footerSecId variable should be declared as ATTRIBUTE_UNUSED:
$ make warn
... [ cut ] ...
cc -c -g -DX86_64 -DLZO -DSNAPPY -DGDB_7_6 qemu-load.c -Wall -O2
-Wstrict-prototypes -Wmissing-prototypes -fstack-protector -Wformat-security
qemu-load.c: In function 'qemu_load':
qemu-load.c:976:13: warning: unused variable 'footerSecId'
[-Wunused-variable]
uint32_t footerSecId = get_be32 (fp);
...
Dave
----- Original Message -----
>
>
> ----- Original Message -----
> > Qemu migration code added new sections to add features
> > for live migration of VM. For loading vmcore file captured
> > with 'virsh dump' we need to parse these sections in crash.
> >
> > This series contains two patches, which parse these sections:
> >
> > patch1: parse 'vm_configuration' section
> > patch2: parse 'vm_footer' section
> >
> > qemu-load.c | 20 ++++++++++++++++++++
> > qemu-load.h | 4 +++-
> > 2 files changed, 23 insertions(+), 1 deletion(-)
> >
> > --
> > Crash-utility mailing list
> > Crash-utility(a)redhat.com
> > https://www.redhat.com/mailman/listinfo/crash-utility
> >
>
> Hi Pankaj,
>
> I suppose I can take the patch. But, given the advent of "virsh dump
--memory-only"
> back in 2012, and more recently, Oleg's added support for QEMU ramdumps by
specifying
> a memory-backend-file object to the qemu-kvm command:
>
> commit 89ed9d0a7f7da4578294a492c1ad857244ce7352
> Author: Dave Anderson <anderson(a)redhat.com>
> Date: Wed May 4 11:50:19 2016 -0400
>
> Introduction of support for "live" ramdump files, such as those that
> are specified by the QEMU mem-path argument of a memory-backend-file
> object. This allows the running of a live crash session against a
> QEMU guest from the host machine. In this example, the /tmp/MEM file
> on a QEMU host represents the guest's physical memory:
>
> $ qemu-kvm ...other-options... \
> -object
> memory-backend-file,id=MEM,size=128m,mem-path=/tmp/MEM,share=on
> \
> -numa node,memdev=MEM -m 128
>
> and a live session run can be run against the guest kernel like so:
>
> $ crash <path-to-guest-vmlinux> live:/tmp/MEM@0
>
> By prepending the ramdump image name with "live:", the crash session
> will
> act as if it were running a normal live session.
> (oleg(a)redhat.com)
>
> And if the guest above crashed, the leftover file can be used as a regular
> ramdump dumpfile:
>
> commit 67a815b8749fbd5c99c29d24b4e699b1d618ddbf
> Author: Dave Anderson <anderson(a)redhat.com>
> Date: Wed May 4 14:34:53 2016 -0400
>
> Added support for x86_64 ramdump files. Without the patch, the crash
> session fails immediately with the message "ramdump: unsupported
> machine type: X86_64".
> (anderson(a)redhat.com)
>
>
> So I don't know why anybody would ever want to use the old "kvmdump"
format?
> It's for all practical purposes deprecated, not to mention that it's really
> an unsuitable excuse for a crash dumpfile.
>
> Dave
>
> --
> 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
2:33 p.m.
New subject: [PATCH 0/2] qemu: Parse necessary sections in crash added in qemu
Thanks Dave.
Best regards,
Pankaj
Hi Pankaj,
Queued for crash-7.1.6:
https://github.com/crash-utility/crash/commit/8cb8e85de65c4b51ca7537186e6...
Dave
----- Original Message -----
>
> BTW, this footerSecId variable should be declared as ATTRIBUTE_UNUSED:
>
> $ make warn
> ... [ cut ] ...
> cc -c -g -DX86_64 -DLZO -DSNAPPY -DGDB_7_6 qemu-load.c -Wall -O2
> -Wstrict-prototypes -Wmissing-prototypes -fstack-protector
> -Wformat-security
> qemu-load.c: In function 'qemu_load':
> qemu-load.c:976:13: warning: unused variable 'footerSecId'
> [-Wunused-variable]
> uint32_t footerSecId = get_be32 (fp);
> ...
>
> Dave
>
>
> ----- Original Message -----
> >
> >
> > ----- Original Message -----
> > > Qemu migration code added new sections to add features
> > > for live migration of VM. For loading vmcore file captured
> > > with 'virsh dump' we need to parse these sections in crash.
> > >
> > > This series contains two patches, which parse these sections:
> > >
> > > patch1: parse 'vm_configuration' section
> > > patch2: parse 'vm_footer' section
> > >
> > > qemu-load.c | 20 ++++++++++++++++++++
> > > qemu-load.h | 4 +++-
> > > 2 files changed, 23 insertions(+), 1 deletion(-)
> > >
> > > --
> > > Crash-utility mailing list
> > > Crash-utility(a)redhat.com
> > > https://www.redhat.com/mailman/listinfo/crash-utility
> > >
> >
> > Hi Pankaj,
> >
> > I suppose I can take the patch. But, given the advent of "virsh dump
> > --memory-only"
> > back in 2012, and more recently, Oleg's added support for QEMU ramdumps
> > by specifying
> > a memory-backend-file object to the qemu-kvm command:
> >
> > commit 89ed9d0a7f7da4578294a492c1ad857244ce7352
> > Author: Dave Anderson <anderson(a)redhat.com>
> > Date: Wed May 4 11:50:19 2016 -0400
> >
> > Introduction of support for "live" ramdump files, such as those
that
> > are specified by the QEMU mem-path argument of a memory-backend-file
> > object. This allows the running of a live crash session against a
> > QEMU guest from the host machine. In this example, the /tmp/MEM file
> > on a QEMU host represents the guest's physical memory:
> >
> > $ qemu-kvm ...other-options... \
> > -object
> > memory-backend-file,id=MEM,size=128m,mem-path=/tmp/MEM,share=on
> > \
> > -numa node,memdev=MEM -m 128
> >
> > and a live session run can be run against the guest kernel like so:
> >
> > $ crash <path-to-guest-vmlinux> live:/tmp/MEM@0
> >
> > By prepending the ramdump image name with "live:", the crash
session
> > will
> > act as if it were running a normal live session.
> > (oleg(a)redhat.com)
> >
> > And if the guest above crashed, the leftover file can be used as a
> > regular
> > ramdump dumpfile:
> >
> > commit 67a815b8749fbd5c99c29d24b4e699b1d618ddbf
> > Author: Dave Anderson <anderson(a)redhat.com>
> > Date: Wed May 4 14:34:53 2016 -0400
> >
> > Added support for x86_64 ramdump files. Without the patch, the crash
> > session fails immediately with the message "ramdump: unsupported
> > machine type: X86_64".
> > (anderson(a)redhat.com)
> >
> >
> > So I don't know why anybody would ever want to use the old
"kvmdump"
> > format?
> > It's for all practical purposes deprecated, not to mention that it's
> > really
> > an unsuitable excuse for a crash dumpfile.
> >
> > Dave
> >
> > --
> > 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
>
--
Crash-utility mailing list
Crash-utility(a)redhat.com
https://www.redhat.com/mailman/listinfo/crash-utility
2970
days inactive
2973
days old
devel@lists.crash-utility.osci.io
10 comments
3 participants
participants (3)
-
Dave Anderson -
Dr. David Alan Gilbert -
Pankaj Gupta