Skip to contentSkip to conversation

vpc: Ignore geometry for large images

RFC PATCH In review Kevin Wolf 5 participants 20 messages Feb 9, 2015 – Feb 18, 2015 Block layer

lore
Share & tools
lore

Kevin Wolf

author
#0 lore raw

RFC PATCH 1/1 vpc: Ignore geometry for large images

The CHS calculation as done per the VHD spec imposes a maximum image size of ~127 GB. Real VHD images exist that are larger than that.

Apparently there are two separate non-standard ways to achieve this:
You could use more heads than the spec does - this is the option that qemu-img create chooses.

However, other images exist where the geometry is set to the maximum (65536/16/255), but the actual image size is larger. Until now, such images are truncated at 127 GB when opening them with qemu.

This patch changes the vpc driver to ignore geometry in this case and only trust the size field in the header.

---

Peter, I'm replacing some of your code in the hope that the new approach is more generally valid. Of course, I haven't tested if your case with disk2vhd is still covered. Could you check this, please?

 block/vpc.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)
block/vpc.c +4 -6
index 46803b1..00409a6 100644
@@ -215,12 +215,10 @@ static int vpc_open(BlockDriverState *bs, QDict *options, int flags,
     bs->total_sectors = (int64_t)
         be16_to_cpu(footer->cyls) * footer->heads * footer->secs_per_cyl;
 
-    /* images created with disk2vhd report a far higher virtual size
-     * than expected with the cyls * heads * sectors_per_cyl formula.
-     * use the footer->size instead if the image was created with
-     * disk2vhd.
-     */
-    if (!strncmp(footer->creator_app, "d2v", 4)) {
+    /* Images that have exactly the maximum geometry are probably bigger and
+     * would be truncated if we adhered to the geometry for them. Rely on
+     * footer->size for them. */
+    if (bs->total_sectors == 65535ULL * 16 * 255) {
         bs->total_sectors = be64_to_cpu(footer->size) / BDRV_SECTOR_SIZE;
     }
 
Signature · 1 line
1.8.3.1
#1 lore raw

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 14 lines

The CHS calculation as done per the VHD spec imposes a maximum image size of ~127 GB. Real VHD images exist that are larger than that.

Apparently there are two separate non-standard ways to achieve this:
You could use more heads than the spec does - this is the option that qemu-img create chooses.

However, other images exist where the geometry is set to the maximum (65536/16/255), but the actual image size is larger. Until now, such images are truncated at 127 GB when opening them with qemu.

This patch changes the vpc driver to ignore geometry in this case and only trust the size field in the header.

Quoted 6 lines

---

Peter, I'm replacing some of your code in the hope that the new approach is more generally valid. Of course, I haven't tested if your case with disk2vhd is still covered. Could you check this, please?

I checked this and found that disk2vhd always sets CHS to 65535ULL * 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.

I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Check vhdCreateImage, vhdOpen in http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

Peter

Quoted 25 lines
  block/vpc.c | 10 ++++------
  1 file changed, 4 insertions(+), 6 deletions(-)
block/vpc.c +4 -6
index 46803b1..00409a6 100644
@@ -215,12 +215,10 @@ static int vpc_open(BlockDriverState *bs, QDict *options, int flags,
      bs->total_sectors = (int64_t)
          be16_to_cpu(footer->cyls) * footer->heads * footer->secs_per_cyl;
 
-    /* images created with disk2vhd report a far higher virtual size
-     * than expected with the cyls * heads * sectors_per_cyl formula.
-     * use the footer->size instead if the image was created with
-     * disk2vhd.
-     */
-    if (!strncmp(footer->creator_app, "d2v", 4)) {
+    /* Images that have exactly the maximum geometry are probably bigger and
+     * would be truncated if we adhered to the geometry for them. Rely on
+     * footer->size for them. */
+    if (bs->total_sectors == 65535ULL * 16 * 255) {
          bs->total_sectors = be64_to_cpu(footer->size) / BDRV_SECTOR_SIZE;
      }
 
Signature · 16 lines
Mit freundlichen Grüßen

Peter Lieven

...........................................................

  KAMP Netzwerkdienste GmbH
  Vestische Str. 89-91 | 46117 Oberhausen
  Tel: +49 (0) 208.89 402-50 | Fax: +49 (0) 208.89 402-40
  [email protected] | http://www.kamp.de

  Geschäftsführer: Heiner Lante | Michael Lante
  Amtsgericht Duisburg | HRB Nr. 12154
  USt-Id-Nr.: DE 120607556

...........................................................
author
#2 lore raw

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 28 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 14 lines

The CHS calculation as done per the VHD spec imposes a maximum image size of ~127 GB. Real VHD images exist that are larger than that.

Apparently there are two separate non-standard ways to achieve this:
You could use more heads than the spec does - this is the option that qemu-img create chooses.

However, other images exist where the geometry is set to the maximum (65536/16/255), but the actual image size is larger. Until now, such images are truncated at 127 GB when opening them with qemu.

This patch changes the vpc driver to ignore geometry in this case and only trust the size field in the header.

Quoted 6 lines

---

Peter, I'm replacing some of your code in the hope that the new approach is more generally valid. Of course, I haven't tested if your case with disk2vhd is still covered. Could you check this, please?

I checked this and found that disk2vhd always sets CHS to 65535ULL * 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.

Quoted 7 lines

I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Check vhdCreateImage, vhdOpen in http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

Kevin

#3 lore raw
Quoted 43 lines · Kevin Wolf

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 36 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 16 lines

The CHS calculation as done per the VHD spec imposes a maximum image size of ~127 GB. Real VHD images exist that are larger than that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the option that qemu-img create chooses.

However, other images exist where the geometry is set to the maximum (65536/16/255), but the actual image size is larger. Until now, such images are truncated at 127 GB when opening them with qemu.

This patch changes the vpc driver to ignore geometry in this case and only trust the size field in the header.

Quoted 6 lines

Peter, I'm replacing some of your code in the hope that the new approach is more generally valid. Of course, I haven't tested if your case with disk2vhd is still covered. Could you check this, please?

I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.

I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Quoted 5 lines

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

Quoted 8 lines

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

author
#4 lore raw

Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 52 lines · Kevin Wolf
Quoted 43 lines

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 36 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 16 lines

The CHS calculation as done per the VHD spec imposes a maximum image size of ~127 GB. Real VHD images exist that are larger than that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the option that qemu-img create chooses.

However, other images exist where the geometry is set to the maximum (65536/16/255), but the actual image size is larger. Until now, such images are truncated at 127 GB when opening them with qemu.

This patch changes the vpc driver to ignore geometry in this case and only trust the size field in the header.

Quoted 6 lines

Peter, I'm replacing some of your code in the hope that the new approach is more generally valid. Of course, I haven't tested if your case with disk2vhd is still covered. Could you check this, please?

I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.

I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Quoted 5 lines

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

Quoted 7 lines

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

Quoted 5 lines

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

Kevin

#5 lore raw

Am 10.02.2015 um 14:54 schrieb Kevin Wolf:

Quoted 55 lines

Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 49 lines · Kevin Wolf
Quoted 41 lines

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 35 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 16 lines

The CHS calculation as done per the VHD spec imposes a maximum image size of ~127 GB. Real VHD images exist that are larger than that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the option that qemu-img create chooses.

However, other images exist where the geometry is set to the maximum (65536/16/255), but the actual image size is larger. Until now, such images are truncated at 127 GB when opening them with qemu.

This patch changes the vpc driver to ignore geometry in this case and only trust the size field in the header.

Quoted 6 lines

Peter, I'm replacing some of your code in the hope that the new approach is more generally valid. Of course, I haven't tested if your case with disk2vhd is still covered. Could you check this, please?

I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.

I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Quoted 5 lines

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

Quoted 7 lines

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

Quoted 5 lines

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

Quoted 7 lines

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

That sounds reasonable. In any case we have to fix qemu-img create to do not create out of spec geometry for images larger than 127G. It should set the correct footer->size and then calculate the geometry.

Peter

#19 lore raw
In reply to Kevin Wolf
Quoted 59 lines · Kevin Wolf

Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 52 lines · Kevin Wolf
Quoted 43 lines

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 36 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 16 lines

The CHS calculation as done per the VHD spec imposes a maximum image size of ~127 GB. Real VHD images exist that are larger than that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the option that qemu-img create chooses.

However, other images exist where the geometry is set to the maximum (65536/16/255), but the actual image size is larger. Until now, such images are truncated at 127 GB when opening them with qemu.

This patch changes the vpc driver to ignore geometry in this case and only trust the size field in the header.

Quoted 6 lines

Peter, I'm replacing some of your code in the hope that the new approach is more generally valid. Of course, I haven't tested if your case with disk2vhd is still covered. Could you check this, please?

I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.

I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Quoted 5 lines

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

Quoted 7 lines

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

Quoted 5 lines

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

Quoted 8 lines

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

Yes, exactly. I don't know if that will catch all the edge use case combinations (an image may have been created by Virtual PC, but formatted and used as a guest under Hyper-V, etc..). But it seems like a reasonable approach that improves the current situation.

author
#6 lore raw
In reply to Peter Lieven

Am 10.02.2015 um 15:00 hat Peter Lieven geschrieben:

Quoted 60 lines

Am 10.02.2015 um 14:54 schrieb Kevin Wolf:

Quoted 59 lines

Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 49 lines · Kevin Wolf
Quoted 41 lines

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 35 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 16 lines

The CHS calculation as done per the VHD spec imposes a maximum image size of ~127 GB. Real VHD images exist that are larger than that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the option that qemu-img create chooses.

However, other images exist where the geometry is set to the maximum (65536/16/255), but the actual image size is larger. Until now, such images are truncated at 127 GB when opening them with qemu.

This patch changes the vpc driver to ignore geometry in this case and only trust the size field in the header.

Quoted 6 lines

Peter, I'm replacing some of your code in the hope that the new approach is more generally valid. Of course, I haven't tested if your case with disk2vhd is still covered. Could you check this, please?

I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.

I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Quoted 5 lines

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

Quoted 7 lines

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

Quoted 5 lines

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Quoted 7 lines
Quoted 3 lines

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

That sounds reasonable. In any case we have to fix qemu-img create to do not create out of spec geometry for images larger than 127G. It should set the correct footer->size and then calculate the geometry.

Do I understand correctly that you just volunteered to fix up that whole thing? ;-)

Kevin

#7 lore raw

Am 10.02.2015 um 15:53 schrieb Kevin Wolf:

Quoted 60 lines

Am 10.02.2015 um 15:00 hat Peter Lieven geschrieben:

Quoted 59 lines

Am 10.02.2015 um 14:54 schrieb Kevin Wolf:

Quoted 58 lines

Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 48 lines · Kevin Wolf
Quoted 40 lines

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 34 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 16 lines

The CHS calculation as done per the VHD spec imposes a maximum image size of ~127 GB. Real VHD images exist that are larger than that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the option that qemu-img create chooses.

However, other images exist where the geometry is set to the maximum (65536/16/255), but the actual image size is larger. Until now, such images are truncated at 127 GB when opening them with qemu.

This patch changes the vpc driver to ignore geometry in this case and only trust the size field in the header.

Quoted 6 lines

Peter, I'm replacing some of your code in the hope that the new approach is more generally valid. Of course, I haven't tested if your case with disk2vhd is still covered. Could you check this, please?

I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.
I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Quoted 5 lines

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

Quoted 7 lines

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

Quoted 5 lines

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Quoted 8 lines
Quoted 6 lines
Quoted 3 lines

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

That sounds reasonable. In any case we have to fix qemu-img create to do not create out of spec geometry for images larger than 127G. It should set the correct footer->size and then calculate the geometry.

Do I understand correctly that you just volunteered to fix up that whole thing? ;-)

I knew that this would happen ;-)

Regarding the C/H/S calculation. I was just wondering if we should not set this to maximum (=invalid?) for all newly created images. That is what disk2vhd does.

Peter

author
#8 lore raw

Am 12.02.2015 um 10:23 hat Peter Lieven geschrieben:

Quoted 67 lines

Am 10.02.2015 um 15:53 schrieb Kevin Wolf:

Quoted 66 lines

Am 10.02.2015 um 15:00 hat Peter Lieven geschrieben:

Quoted 65 lines

Am 10.02.2015 um 14:54 schrieb Kevin Wolf:

Quoted 54 lines

Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 48 lines · Kevin Wolf
Quoted 40 lines

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 34 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 16 lines

The CHS calculation as done per the VHD spec imposes a maximum image size of ~127 GB. Real VHD images exist that are larger than that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the option that qemu-img create chooses.

However, other images exist where the geometry is set to the maximum (65536/16/255), but the actual image size is larger. Until now, such images are truncated at 127 GB when opening them with qemu.

This patch changes the vpc driver to ignore geometry in this case and only trust the size field in the header.

Quoted 6 lines

Peter, I'm replacing some of your code in the hope that the new approach is more generally valid. Of course, I haven't tested if your case with disk2vhd is still covered. Could you check this, please?

I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.
I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Quoted 5 lines

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

Quoted 7 lines

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

Quoted 5 lines

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

Quoted 7 lines

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

That sounds reasonable. In any case we have to fix qemu-img create to do not create out of spec geometry for images larger than 127G. It should set the correct footer->size and then calculate the geometry.

Quoted 8 lines
Quoted 2 lines

Do I understand correctly that you just volunteered to fix up that whole thing? ;-)

I knew that this would happen ;-)

Regarding the C/H/S calculation. I was just wondering if we should not set this to maximum (=invalid?) for all newly created images. That is what disk2vhd does.

CHS is what Virtual PC relies on. So I guess if you did that, you would render images unusable by it. Are you sure that disk2vhd does this always? I would have thought that it only does it for large images.

Kevin

#9 lore raw

Am 12.02.2015 um 10:58 schrieb Kevin Wolf:

Quoted 70 lines

Am 12.02.2015 um 10:23 hat Peter Lieven geschrieben:

Quoted 69 lines

Am 10.02.2015 um 15:53 schrieb Kevin Wolf:

Quoted 68 lines

Am 10.02.2015 um 15:00 hat Peter Lieven geschrieben:

Quoted 59 lines

Am 10.02.2015 um 14:54 schrieb Kevin Wolf:

Quoted 58 lines

Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 48 lines · Kevin Wolf
Quoted 40 lines

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 34 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 16 lines

The CHS calculation as done per the VHD spec imposes a maximum image size of ~127 GB. Real VHD images exist that are larger than that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the option that qemu-img create chooses.

However, other images exist where the geometry is set to the maximum (65536/16/255), but the actual image size is larger. Until now, such images are truncated at 127 GB when opening them with qemu.

This patch changes the vpc driver to ignore geometry in this case and only trust the size field in the header.

Quoted 6 lines

Peter, I'm replacing some of your code in the hope that the new approach is more generally valid. Of course, I haven't tested if your case with disk2vhd is still covered. Could you check this, please?

I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.
I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Quoted 5 lines

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

Quoted 7 lines

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

Quoted 5 lines

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Quoted 6 lines
Quoted 3 lines

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

That sounds reasonable. In any case we have to fix qemu-img create to do not create out of spec geometry for images larger than 127G. It should set the correct footer->size and then calculate the geometry.

Do I understand correctly that you just volunteered to fix up that whole thing? ;-)

Quoted 8 lines
Quoted 5 lines

I knew that this would happen ;-)

Regarding the C/H/S calculation. I was just wondering if we should not set this to maximum (=invalid?) for all newly created images. That is what disk2vhd does.

CHS is what Virtual PC relies on. So I guess if you did that, you would render images unusable by it. Are you sure that disk2vhd does this always? I would have thought that it only does it for large images.

At least 2.0.1 (latest available version) does this as well as the version that I used when I added the hack for d2v creator.

Virtual PC would not be able to use images we create with qemu-img create if we use footer->size (which I suppose to reanme to footer->cur_size, btw) to calculate bs->total_sectors because we might write data to the end of the image which gets truncated in CHS format.

Peter

author
#10 lore raw

Am 12.02.2015 um 11:02 hat Peter Lieven geschrieben:

Quoted 80 lines

Am 12.02.2015 um 10:58 schrieb Kevin Wolf:

Quoted 70 lines

Am 12.02.2015 um 10:23 hat Peter Lieven geschrieben:

Quoted 69 lines

Am 10.02.2015 um 15:53 schrieb Kevin Wolf:

Quoted 68 lines

Am 10.02.2015 um 15:00 hat Peter Lieven geschrieben:

Quoted 59 lines

Am 10.02.2015 um 14:54 schrieb Kevin Wolf:

Quoted 58 lines

Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 48 lines · Kevin Wolf
Quoted 40 lines

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 34 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 16 lines

The CHS calculation as done per the VHD spec imposes a maximum image size of ~127 GB. Real VHD images exist that are larger than that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the option that qemu-img create chooses.

However, other images exist where the geometry is set to the maximum (65536/16/255), but the actual image size is larger. Until now, such images are truncated at 127 GB when opening them with qemu.

This patch changes the vpc driver to ignore geometry in this case and only trust the size field in the header.

Quoted 6 lines

Peter, I'm replacing some of your code in the hope that the new approach is more generally valid. Of course, I haven't tested if your case with disk2vhd is still covered. Could you check this, please?

I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.
I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Quoted 5 lines

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

Quoted 7 lines

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

Quoted 5 lines

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Quoted 6 lines
Quoted 3 lines

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

That sounds reasonable. In any case we have to fix qemu-img create to do not create out of spec geometry for images larger than 127G. It should set the correct footer->size and then calculate the geometry.

Do I understand correctly that you just volunteered to fix up that whole thing? ;-)

Quoted 8 lines
Quoted 5 lines

I knew that this would happen ;-)

Regarding the C/H/S calculation. I was just wondering if we should not set this to maximum (=invalid?) for all newly created images. That is what disk2vhd does.

CHS is what Virtual PC relies on. So I guess if you did that, you would render images unusable by it. Are you sure that disk2vhd does this always? I would have thought that it only does it for large images.

Quoted 7 lines

At least 2.0.1 (latest available version) does this as well as the version that I used when I added the hack for d2v creator.

Virtual PC would not be able to use images we create with qemu-img create if we use footer->size (which I suppose to reanme to footer->cur_size, btw) to calculate bs->total_sectors because we might write data to the end of the image which gets truncated in CHS format.

These kinds of problems are why I'd like to keep CHS and size always consistent when creating an image with qemu-img.

Kevin

#11 lore raw

Am 12.02.2015 um 11:06 schrieb Kevin Wolf:

Quoted 83 lines

Am 12.02.2015 um 11:02 hat Peter Lieven geschrieben:

Quoted 82 lines

Am 12.02.2015 um 10:58 schrieb Kevin Wolf:

Quoted 70 lines

Am 12.02.2015 um 10:23 hat Peter Lieven geschrieben:

Quoted 69 lines

Am 10.02.2015 um 15:53 schrieb Kevin Wolf:

Quoted 68 lines

Am 10.02.2015 um 15:00 hat Peter Lieven geschrieben:

Quoted 59 lines

Am 10.02.2015 um 14:54 schrieb Kevin Wolf:

Quoted 58 lines

Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 48 lines · Kevin Wolf
Quoted 40 lines

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 34 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 16 lines

The CHS calculation as done per the VHD spec imposes a maximum image size of ~127 GB. Real VHD images exist that are larger than that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the option that qemu-img create chooses.

However, other images exist where the geometry is set to the maximum (65536/16/255), but the actual image size is larger. Until now, such images are truncated at 127 GB when opening them with qemu.

This patch changes the vpc driver to ignore geometry in this case and only trust the size field in the header.

Quoted 6 lines

Peter, I'm replacing some of your code in the hope that the new approach is more generally valid. Of course, I haven't tested if your case with disk2vhd is still covered. Could you check this, please?

I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.
I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Quoted 5 lines

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

Quoted 7 lines

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

Quoted 5 lines

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Quoted 6 lines
Quoted 3 lines

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

That sounds reasonable. In any case we have to fix qemu-img create to do not create out of spec geometry for images larger than 127G. It should set the correct footer->size and then calculate the geometry.

Do I understand correctly that you just volunteered to fix up that whole thing? ;-)

Quoted 8 lines
Quoted 5 lines

I knew that this would happen ;-)

Regarding the C/H/S calculation. I was just wondering if we should not set this to maximum (=invalid?) for all newly created images. That is what disk2vhd does.

CHS is what Virtual PC relies on. So I guess if you did that, you would render images unusable by it. Are you sure that disk2vhd does this always? I would have thought that it only does it for large images.

At least 2.0.1 (latest available version) does this as well as the version that I used when I added the hack for d2v creator.

Quoted 6 lines
Quoted 4 lines

Virtual PC would not be able to use images we create with qemu-img create if we use footer->size (which I suppose to reanme to footer->cur_size, btw) to calculate bs->total_sectors because we might write data to the end of the image which gets truncated in CHS format.

These kinds of problems are why I'd like to keep CHS and size always consistent when creating an image with qemu-img.

Okay, then I would vote for your RFC patch + fixing qemu-img create to not generate out of spec CHS values and just set maximum which then would make vpc_open use footer->size.

Peter

Quoted 2 lines

Kevin

Signature · 16 lines
Mit freundlichen Grüßen

Peter Lieven

...........................................................

  KAMP Netzwerkdienste GmbH
  Vestische Str. 89-91 | 46117 Oberhausen
  Tel: +49 (0) 208.89 402-50 | Fax: +49 (0) 208.89 402-40
  [email protected] | http://www.kamp.de

  Geschäftsführer: Heiner Lante | Michael Lante
  Amtsgericht Duisburg | HRB Nr. 12154
  USt-Id-Nr.: DE 120607556

...........................................................
author
#12 lore raw

Am 12.02.2015 um 11:09 hat Peter Lieven geschrieben:

Quoted 88 lines

Am 12.02.2015 um 11:06 schrieb Kevin Wolf:

Quoted 87 lines

Am 12.02.2015 um 11:02 hat Peter Lieven geschrieben:

Quoted 86 lines

Am 12.02.2015 um 10:58 schrieb Kevin Wolf:

Quoted 70 lines

Am 12.02.2015 um 10:23 hat Peter Lieven geschrieben:

Quoted 69 lines

Am 10.02.2015 um 15:53 schrieb Kevin Wolf:

Quoted 68 lines

Am 10.02.2015 um 15:00 hat Peter Lieven geschrieben:

Quoted 59 lines

Am 10.02.2015 um 14:54 schrieb Kevin Wolf:

Quoted 58 lines

Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 48 lines · Kevin Wolf
Quoted 40 lines

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 34 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 16 lines

The CHS calculation as done per the VHD spec imposes a maximum image size of ~127 GB. Real VHD images exist that are larger than that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the option that qemu-img create chooses.

However, other images exist where the geometry is set to the maximum (65536/16/255), but the actual image size is larger. Until now, such images are truncated at 127 GB when opening them with qemu.

This patch changes the vpc driver to ignore geometry in this case and only trust the size field in the header.

Quoted 6 lines

Peter, I'm replacing some of your code in the hope that the new approach is more generally valid. Of course, I haven't tested if your case with disk2vhd is still covered. Could you check this, please?

I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.
I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Quoted 5 lines

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

Quoted 7 lines

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

Quoted 5 lines

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Quoted 6 lines
Quoted 3 lines

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

That sounds reasonable. In any case we have to fix qemu-img create to do not create out of spec geometry for images larger than 127G. It should set the correct footer->size and then calculate the geometry.

Do I understand correctly that you just volunteered to fix up that whole thing? ;-)

Quoted 8 lines
Quoted 5 lines

I knew that this would happen ;-)

Regarding the C/H/S calculation. I was just wondering if we should not set this to maximum (=invalid?) for all newly created images. That is what disk2vhd does.

CHS is what Virtual PC relies on. So I guess if you did that, you would render images unusable by it. Are you sure that disk2vhd does this always? I would have thought that it only does it for large images.

At least 2.0.1 (latest available version) does this as well as the version that I used when I added the hack for d2v creator.

Virtual PC would not be able to use images we create with qemu-img create if we use footer->size (which I suppose to reanme to footer->cur_size, btw) to calculate bs->total_sectors because we might write data to the end of the image which gets truncated in CHS format.

Quoted 6 lines
Quoted 2 lines

These kinds of problems are why I'd like to keep CHS and size always consistent when creating an image with qemu-img.

Okay, then I would vote for your RFC patch + fixing qemu-img create to not generate out of spec CHS values and just set maximum which then would make vpc_open use footer->size.

Really the RFC patch or what we discussed above ("vpc" creator = CHS, everything else = footer->size)? Once I know what we prefer, I'll send the real patch.

As for heads > 16, that would essentially mean reverting 258d2edb. Should be easy to do, the harder part is probably the commit message explaining why it's helpful and safe. Note that the commit message of 258d2edb claims that it's not out of spec. I _think_ we can do the revert with a good explanation, but I'll leave that to you.

(CCed Charles who wrote that commit)

Kevin

#13 lore raw

Am 12.02.2015 um 11:23 schrieb Kevin Wolf:

Quoted 89 lines

Am 12.02.2015 um 11:09 hat Peter Lieven geschrieben:

Quoted 88 lines

Am 12.02.2015 um 11:06 schrieb Kevin Wolf:

Quoted 87 lines

Am 12.02.2015 um 11:02 hat Peter Lieven geschrieben:

Quoted 86 lines

Am 12.02.2015 um 10:58 schrieb Kevin Wolf:

Quoted 70 lines

Am 12.02.2015 um 10:23 hat Peter Lieven geschrieben:

Quoted 69 lines

Am 10.02.2015 um 15:53 schrieb Kevin Wolf:

Quoted 68 lines

Am 10.02.2015 um 15:00 hat Peter Lieven geschrieben:

Quoted 59 lines

Am 10.02.2015 um 14:54 schrieb Kevin Wolf:

Quoted 58 lines

Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 48 lines · Kevin Wolf
Quoted 40 lines

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 34 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 22 lines

The CHS calculation as done per the VHD spec imposes a maximum
image size of ~127 GB. Real VHD images exist that are larger than
that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the
option that qemu-img create chooses.

However, other images exist where the geometry is set to the
maximum (65536/16/255), but the actual image size is larger.
Until now, such images are truncated at 127 GB when opening them
with qemu.

This patch changes the vpc driver to ignore geometry in this case
and only trust the size field in the header.

Signed-off-by: Kevin Wolf <[email protected]> ---

Peter, I'm replacing some of your code in the hope that the new
approach is more generally valid. Of course, I haven't tested if
your case with disk2vhd is still covered. Could you check this,
please?

I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.
I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Quoted 5 lines

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

Quoted 7 lines

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

Quoted 5 lines

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Quoted 6 lines
Quoted 3 lines

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

That sounds reasonable. In any case we have to fix qemu-img create to do not create out of spec geometry for images larger than 127G. It should set the correct footer->size and then calculate the geometry.

Do I understand correctly that you just volunteered to fix up that whole thing? ;-)

Quoted 8 lines
Quoted 5 lines

I knew that this would happen ;-)

Regarding the C/H/S calculation. I was just wondering if we should not set this to maximum (=invalid?) for all newly created images. That is what disk2vhd does.

CHS is what Virtual PC relies on. So I guess if you did that, you would render images unusable by it. Are you sure that disk2vhd does this always? I would have thought that it only does it for large images.

At least 2.0.1 (latest available version) does this as well as the version that I used when I added the hack for d2v creator.

Virtual PC would not be able to use images we create with qemu-img create if we use footer->size (which I suppose to reanme to footer->cur_size, btw) to calculate bs->total_sectors because we might write data to the end of the image which gets truncated in CHS format.

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

These kinds of problems are why I'd like to keep CHS and size always consistent when creating an image with qemu-img.

Okay, then I would vote for your RFC patch + fixing qemu-img create to not generate out of spec CHS values and just set maximum which then would make vpc_open use footer->size.

Really the RFC patch or what we discussed above ("vpc" creator = CHS, everything else = footer->size)? Once I know what we prefer, I'll send the real patch.

The RFC patch as is.

Quoted 6 lines

As for heads > 16, that would essentially mean reverting 258d2edb. Should be easy to do, the harder part is probably the commit message explaining why it's helpful and safe. Note that the commit message of 258d2edb claims that it's not out of spec. I _think_ we can do the revert with a good explanation, but I'll leave that to you.

The only spec I have access to is [1].
If you check the CHS calculation code in the appendix, you will find:

if (totalSectors > 65535 * 16 * 255)

{

totalSectors = 65535 * 16 * 255;

}

[1] http://download.microsoft.com/download/f/f/e/ffef50a5-07dd-4cf8-aaa3-442c0673a029/Virtual%20Hard%20Disk%20Format%20Spec_10_18_06.doc

Peter

#14 lore raw
In reply to Kevin Wolf
Quoted 101 lines
Quoted 1 line
Quoted 1 line · Kevin Wolf

Am 12.02.2015 um 11:09 hat Peter Lieven geschrieben:

Quoted 88 lines

Am 12.02.2015 um 11:06 schrieb Kevin Wolf:

Quoted 87 lines

Am 12.02.2015 um 11:02 hat Peter Lieven geschrieben:

Quoted 86 lines

Am 12.02.2015 um 10:58 schrieb Kevin Wolf:

Quoted 70 lines

Am 12.02.2015 um 10:23 hat Peter Lieven geschrieben:

Quoted 69 lines

Am 10.02.2015 um 15:53 schrieb Kevin Wolf:

Quoted 68 lines

Am 10.02.2015 um 15:00 hat Peter Lieven geschrieben:

Quoted 59 lines

Am 10.02.2015 um 14:54 schrieb Kevin Wolf:

Quoted 58 lines

Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 48 lines · Kevin Wolf
Quoted 40 lines

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 34 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:

Quoted 22 lines

The CHS calculation as done per the VHD spec imposes a maximum
image size of ~127 GB. Real VHD images exist that are larger than
that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the
option that qemu-img create chooses.

However, other images exist where the geometry is set to the
maximum (65536/16/255), but the actual image size is larger.
Until now, such images are truncated at 127 GB when opening them
with qemu.

This patch changes the vpc driver to ignore geometry in this case
and only trust the size field in the header.

Signed-off-by: Kevin Wolf <[email protected]> ---

Peter, I'm replacing some of your code in the hope that the new
approach is more generally valid. Of course, I haven't tested if
your case with disk2vhd is still covered. Could you check this,
please?

I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best solution to ignore CHS completely and always derive total_sectors from footer->size unconditionally.
I had a look at what virtualbox does and they only rely on footer->size. If they alter the size or create an image the write the new size into the footer and recalculate CHS by the formula found in the appendix of the original spec.

Quoted 5 lines

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in an ATA controller only.

Quoted 7 lines

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

Quoted 5 lines

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Quoted 6 lines
Quoted 3 lines

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

That sounds reasonable. In any case we have to fix qemu-img create to do not create out of spec geometry for images larger than 127G. It should set the correct footer->size and then calculate the geometry.

Do I understand correctly that you just volunteered to fix up that whole thing? ;-)

Quoted 8 lines
Quoted 5 lines

I knew that this would happen ;-)

Regarding the C/H/S calculation. I was just wondering if we should not set this to maximum (=invalid?) for all newly created images. That is what disk2vhd does.

CHS is what Virtual PC relies on. So I guess if you did that, you would render images unusable by it. Are you sure that disk2vhd does this always? I would have thought that it only does it for large images.

At least 2.0.1 (latest available version) does this as well as the version that I used when I added the hack for d2v creator.

Virtual PC would not be able to use images we create with qemu-img create if we use footer->size (which I suppose to reanme to footer->cur_size, btw) to calculate bs->total_sectors because we might write data to the end of the image which gets truncated in CHS format.

Quoted 6 lines
Quoted 2 lines

These kinds of problems are why I'd like to keep CHS and size always consistent when creating an image with qemu-img.

Okay, then I would vote for your RFC patch + fixing qemu-img create to not generate out of spec CHS values and just set maximum which then would make vpc_open use footer->size.

Really the RFC patch or what we discussed above ("vpc" creator = CHS, everything else = footer->size)? Once I know what we prefer, I'll send the real patch.

Quoted 7 lines

As for heads > 16, that would essentially mean reverting 258d2edb. Should be easy to do, the harder part is probably the commit message explaining why it's helpful and safe. Note that the commit message of 258d2edb claims that it's not out of spec. I _think_ we can do the revert with a good explanation, but I'll leave that to you.

(CCed Charles who wrote that commit)

IIUC, the plan is to revert my old commit and use the footer->size field to describe images greater than 127 GB. This change would break other tools from Virtual PC, Xens vhd-util and maybe others from reading images greater than 127 GB because the head field would be forced back to using 16 and these tools won't know to check the footer->size field. Is there any reason not to keep the original commit and still use the footer->size field?

A purist would argue that heads must be 16 for true ATA emulation but allowing up to 255 doesn't seem to matter and the VHD spec does support up to 2 TB.

- Charles

#15 lore raw

Am 12.02.2015 um 18:18 schrieb Charles Arnold <[email protected]>:

Quoted 107 lines
Quoted 107 lines
Quoted 1 line
Quoted 1 line · Kevin Wolf

Am 12.02.2015 um 11:09 hat Peter Lieven geschrieben:

Quoted 88 lines
Quoted 88 lines

Am 12.02.2015 um 11:06 schrieb Kevin Wolf:
Am 12.02.2015 um 11:02 hat Peter Lieven geschrieben:

Quoted 86 lines
Quoted 71 lines

Am 12.02.2015 um 10:58 schrieb Kevin Wolf:
Am 12.02.2015 um 10:23 hat Peter Lieven geschrieben:

Quoted 69 lines
Quoted 69 lines

Am 10.02.2015 um 15:53 schrieb Kevin Wolf:
Am 10.02.2015 um 15:00 hat Peter Lieven geschrieben:

Quoted 59 lines
Quoted 59 lines

Am 10.02.2015 um 14:54 schrieb Kevin Wolf:
Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 48 lines · Kevin Wolf
Quoted 40 lines

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:

Quoted 39 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:
The CHS calculation as done per the VHD spec imposes a maximum
image size of ~127 GB. Real VHD images exist that are larger than
that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the
option that qemu-img create chooses.

However, other images exist where the geometry is set to the
maximum (65536/16/255), but the actual image size is larger.
Until now, such images are truncated at 127 GB when opening them
with qemu.

This patch changes the vpc driver to ignore geometry in this case
and only trust the size field in the header.

Signed-off-by: Kevin Wolf <[email protected]> ---

Peter, I'm replacing some of your code in the hope that the new
approach is more generally valid. Of course, I haven't tested if
your case with disk2vhd is still covered. Could you check this,
please?
I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best
solution to ignore CHS completely and always derive total_sectors
from footer->size unconditionally.
I had a look at what virtualbox does and they only rely on
footer->size. If they alter the size or create an image the write
the new size into the footer and recalculate CHS by the formula
found in the appendix of the original spec.

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in
an ATA controller only.

Quoted 7 lines

The problem with just using footer->size back then when I implemented this was that from the perspective of a VirtualPC guest run in qemu, the size of its hard disk would change, which you don't want either. Going from VPC to qemu would be ugly, but mostly harmless as the disk only grows. But if you use an image in qemu where the disk looks larger and then go back to VPC which respects geometry, your data may be truncated.

Quoted 5 lines

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Quoted 6 lines
Quoted 3 lines

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

That sounds reasonable. In any case we have to fix qemu-img create to do not create out of spec geometry for images larger than 127G. It should set the correct footer->size and then calculate the geometry.

Do I understand correctly that you just volunteered to fix up that whole thing? ;-)

Quoted 8 lines
Quoted 5 lines

I knew that this would happen ;-)

Regarding the C/H/S calculation. I was just wondering if we should not set this to maximum (=invalid?) for all newly created images. That is what disk2vhd does.

CHS is what Virtual PC relies on. So I guess if you did that, you would render images unusable by it. Are you sure that disk2vhd does this always? I would have thought that it only does it for large images.

At least 2.0.1 (latest available version) does this as well as the version that I used when I added the hack for d2v creator.

Virtual PC would not be able to use images we create with qemu-img create if we use footer->size (which I suppose to reanme to footer->cur_size, btw) to calculate bs->total_sectors because we might write data to the end of the image which gets truncated in CHS format.

Quoted 6 lines
Quoted 2 lines

These kinds of problems are why I'd like to keep CHS and size always consistent when creating an image with qemu-img.

Okay, then I would vote for your RFC patch + fixing qemu-img create to not generate out of spec CHS values and just set maximum which then would make vpc_open use footer->size.

Really the RFC patch or what we discussed above ("vpc" creator = CHS, everything else = footer->size)? Once I know what we prefer, I'll send the real patch.

As for heads > 16, that would essentially mean reverting 258d2edb. Should be easy to do, the harder part is probably the commit message explaining why it's helpful and safe. Note that the commit message of 258d2edb claims that it's not out of spec. I _think_ we can do the revert with a good explanation, but I'll leave that to you.

Quoted 8 lines
Quoted 1 line

(CCed Charles who wrote that commit)

IIUC, the plan is to revert my old commit and use the footer->size field to describe images greater than 127 GB. This change would break other tools from Virtual PC, Xens vhd-util and maybe others from reading images greater than 127 GB because the head field would be forced back to using 16 and these tools won't know to check the footer->size field. Is there any reason not to keep the original commit and still use the footer->size field?

do you have a Pointer to a spec that is newer than 2006? the one i have describes CHS calculation up to 65535 x 16 x 255 sectors. that is set as Maximum if total sectors is higher. I would do the same when writing a footer. in vhd_open I would derive total_sectors from C x H x S except for the case that it is exactly 65535 x 16 x 255. In this case I would take footer->size / 512. Virtualbox does it that way and at the comment from Stefan in the commit message for your Patch suggest that you observed a similar behaviour for HyperV.

Peter

Quoted 6 lines

A purist would argue that heads must be 16 for true ATA emulation but allowing up to 255 doesn't seem to matter and the VHD spec does support up to 2 TB.

- Charles

#16 lore raw
Quoted 113 lines
Quoted 1 line
Quoted 1 line · Peter Lieven

Am 12.02.2015 um 18:18 schrieb Charles Arnold <[email protected]>:

Quoted 109 lines
Quoted 108 lines
Quoted 1 line
Quoted 1 line · Kevin Wolf

Am 12.02.2015 um 11:09 hat Peter Lieven geschrieben:

Quoted 88 lines
Quoted 88 lines

Am 12.02.2015 um 11:06 schrieb Kevin Wolf:
Am 12.02.2015 um 11:02 hat Peter Lieven geschrieben:

Quoted 86 lines
Quoted 71 lines

Am 12.02.2015 um 10:58 schrieb Kevin Wolf:
Am 12.02.2015 um 10:23 hat Peter Lieven geschrieben:

Quoted 69 lines
Quoted 69 lines

Am 10.02.2015 um 15:53 schrieb Kevin Wolf:
Am 10.02.2015 um 15:00 hat Peter Lieven geschrieben:

Quoted 59 lines
Quoted 59 lines

Am 10.02.2015 um 14:54 schrieb Kevin Wolf:
Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 48 lines · Kevin Wolf
Quoted 47 lines

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:
Am 09.02.2015 um 17:09 schrieb Kevin Wolf:
The CHS calculation as done per the VHD spec imposes a maximum
image size of ~127 GB. Real VHD images exist that are larger than
that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the
option that qemu-img create chooses.

However, other images exist where the geometry is set to the
maximum (65536/16/255), but the actual image size is larger.
Until now, such images are truncated at 127 GB when opening them
with qemu.

This patch changes the vpc driver to ignore geometry in this case
and only trust the size field in the header.

Signed-off-by: Kevin Wolf <[email protected]> ---

Peter, I'm replacing some of your code in the hope that the new
approach is more generally valid. Of course, I haven't tested if
your case with disk2vhd is still covered. Could you check this,
please?
I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best
solution to ignore CHS completely and always derive total_sectors
from footer->size unconditionally.
I had a look at what virtualbox does and they only rely on
footer->size. If they alter the size or create an image the write
the new size into the footer and recalculate CHS by the formula
found in the appendix of the original spec.

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in
an ATA controller only.
The problem with just using footer->size back then when I
implemented this was that from the perspective of a VirtualPC guest
run in qemu, the size of its hard disk would change, which you don't
want either. Going from VPC to qemu would be ugly, but mostly
harmless as the disk only grows. But if you use an image in qemu
where the disk looks larger and then go back to VPC which respects
geometry, your data may be truncated.

Quoted 5 lines

I believe the vpc "creator" field is different if the image was created by Virtual PC, versus created by Hyper-V ("vpc" and "win", respectively, I think). Perhaps we could use that to infer a guest image came from VirtualPC, and thus not use footer->size in that scenario?

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Quoted 6 lines
Quoted 3 lines

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

That sounds reasonable. In any case we have to fix qemu-img create to do not create out of spec geometry for images larger than 127G. It should set the correct footer->size and then calculate the geometry.

Do I understand correctly that you just volunteered to fix up that whole thing? ;-)

Quoted 8 lines
Quoted 5 lines

I knew that this would happen ;-)

Regarding the C/H/S calculation. I was just wondering if we should not set this to maximum (=invalid?) for all newly created images. That is what disk2vhd does.

CHS is what Virtual PC relies on. So I guess if you did that, you would render images unusable by it. Are you sure that disk2vhd does this always? I would have thought that it only does it for large images.

At least 2.0.1 (latest available version) does this as well as the version that I used when I added the hack for d2v creator.

Virtual PC would not be able to use images we create with qemu-img create if we use footer->size (which I suppose to reanme to footer->cur_size, btw) to calculate bs->total_sectors because we might write data to the end of the image which gets truncated in CHS format.

Quoted 6 lines
Quoted 2 lines

These kinds of problems are why I'd like to keep CHS and size always consistent when creating an image with qemu-img.

Okay, then I would vote for your RFC patch + fixing qemu-img create to not generate out of spec CHS values and just set maximum which then would make vpc_open use footer->size.

Really the RFC patch or what we discussed above ("vpc" creator = CHS, everything else = footer->size)? Once I know what we prefer, I'll send the real patch.

As for heads > 16, that would essentially mean reverting 258d2edb. Should be easy to do, the harder part is probably the commit message explaining why it's helpful and safe. Note that the commit message of 258d2edb claims that it's not out of spec. I _think_ we can do the revert with a good explanation, but I'll leave that to you.

(CCed Charles who wrote that commit)

Quoted 8 lines
Quoted 6 lines

IIUC, the plan is to revert my old commit and use the footer->size field to describe images greater than 127 GB. This change would break other tools from Virtual PC, Xens vhd-util and maybe others from reading images greater than 127 GB because the head field would be forced back to using 16 and these tools won't know to check the footer->size field. Is there any reason not to keep the original commit and still use the footer->size field?

do you have a Pointer to a spec that is newer than 2006?

No, that is the most recent.

Quoted 10 lines

the one i have
describes CHS calculation up to 65535 x 16 x 255 sectors. that is set as Maximum if total sectors is higher.
I would do the same when writing a
footer. in vhd_open I would derive total_sectors from C x H x S except for the case that it is exactly 65535 x 16 x 255. In this case I would take footer->size / 512.
Virtualbox does it that way and at the comment from Stefan in the commit message for your Patch suggest that you observed a similar behaviour for HyperV.

Right, although this was a long time ago for me to remember the specifics :) In the end I think supporting at least the 2 TB size allowed by the spec is what we need while not breaking existing images.

- Charles

#17 lore raw

Am 12.02.2015 um 22:01 schrieb Charles Arnold:

Quoted 128 lines
Quoted 108 lines
Quoted 1 line
Quoted 1 line · Peter Lieven

Am 12.02.2015 um 18:18 schrieb Charles Arnold <[email protected]>:

Quoted 105 lines
Quoted 105 lines
Quoted 1 line
Quoted 1 line · Kevin Wolf

Am 12.02.2015 um 11:09 hat Peter Lieven geschrieben:

Quoted 88 lines
Quoted 88 lines

Am 12.02.2015 um 11:06 schrieb Kevin Wolf:
Am 12.02.2015 um 11:02 hat Peter Lieven geschrieben:

Quoted 86 lines
Quoted 71 lines

Am 12.02.2015 um 10:58 schrieb Kevin Wolf:
Am 12.02.2015 um 10:23 hat Peter Lieven geschrieben:

Quoted 69 lines
Quoted 69 lines

Am 10.02.2015 um 15:53 schrieb Kevin Wolf:
Am 10.02.2015 um 15:00 hat Peter Lieven geschrieben:

Quoted 59 lines
Quoted 59 lines

Am 10.02.2015 um 14:54 schrieb Kevin Wolf:
Am 10.02.2015 um 14:42 hat Jeff Cody geschrieben:

Quoted 53 lines · Kevin Wolf

Am 10.02.2015 um 12:41 hat Peter Lieven geschrieben:
Am 09.02.2015 um 17:09 schrieb Kevin Wolf:
The CHS calculation as done per the VHD spec imposes a maximum
image size of ~127 GB. Real VHD images exist that are larger than
that.

Apparently there are two separate non-standard ways to achieve
this: You could use more heads than the spec does - this is the
option that qemu-img create chooses.

However, other images exist where the geometry is set to the
maximum (65536/16/255), but the actual image size is larger.
Until now, such images are truncated at 127 GB when opening them
with qemu.

This patch changes the vpc driver to ignore geometry in this case
and only trust the size field in the header.

Signed-off-by: Kevin Wolf <[email protected]> ---

Peter, I'm replacing some of your code in the hope that the new
approach is more generally valid. Of course, I haven't tested if
your case with disk2vhd is still covered. Could you check this,
please?
I checked this and found that disk2vhd always sets CHS to 65535ULL
* 16 * 255 independed of the real size.

But, as the conversion to CHS may have an error its maybe the best
solution to ignore CHS completely and always derive total_sectors
from footer->size unconditionally.
I had a look at what virtualbox does and they only rely on
footer->size. If they alter the size or create an image the write
the new size into the footer and recalculate CHS by the formula
found in the appendix of the original spec.

Check vhdCreateImage, vhdOpen in
http://www.virtualbox.org/svn/vbox/trunk/src/VBox/Storage/VHD.cpp

The original spec also says that CHS values purpose is the use in
an ATA controller only.
The problem with just using footer->size back then when I
implemented this was that from the perspective of a VirtualPC guest
run in qemu, the size of its hard disk would change, which you don't
want either. Going from VPC to qemu would be ugly, but mostly
harmless as the disk only grows. But if you use an image in qemu
where the disk looks larger and then go back to VPC which respects
geometry, your data may be truncated.
I believe the vpc "creator" field is different if the image was
created by Virtual PC, versus created by Hyper-V ("vpc" and "win",
respectively, I think). Perhaps we could use that to infer a guest
image came from VirtualPC, and thus not use footer->size in that
scenario?

Right, I think we discussed that before. Do you remember the outcome of that discussion? I seem to remember that we had a conclusion, but apparently it was never actually implemented.

Quoted 6 lines
Quoted 3 lines

Would your proposal be to special-case "vpc" to apply the geometry, and everything else (including "win", "d2v" and "qemu") would use the footer field?

That sounds reasonable. In any case we have to fix qemu-img create to do not create out of spec geometry for images larger than 127G. It should set the correct footer->size and then calculate the geometry.

Do I understand correctly that you just volunteered to fix up that whole thing? ;-)

Quoted 8 lines
Quoted 5 lines

I knew that this would happen ;-)

Regarding the C/H/S calculation. I was just wondering if we should not set this to maximum (=invalid?) for all newly created images. That is what disk2vhd does.

CHS is what Virtual PC relies on. So I guess if you did that, you would render images unusable by it. Are you sure that disk2vhd does this always? I would have thought that it only does it for large images.

At least 2.0.1 (latest available version) does this as well as the version that I used when I added the hack for d2v creator.

Virtual PC would not be able to use images we create with qemu-img create if we use footer->size (which I suppose to reanme to footer->cur_size, btw) to calculate bs->total_sectors because we might write data to the end of the image which gets truncated in CHS format.

Quoted 5 lines
Quoted 2 lines

These kinds of problems are why I'd like to keep CHS and size always consistent when creating an image with qemu-img.

Okay, then I would vote for your RFC patch + fixing qemu-img create to not generate out of spec CHS values and just set maximum which then would make vpc_open use footer->size.

Really the RFC patch or what we discussed above ("vpc" creator = CHS, everything else = footer->size)? Once I know what we prefer, I'll send the real patch.

As for heads > 16, that would essentially mean reverting 258d2edb. Should be easy to do, the harder part is probably the commit message explaining why it's helpful and safe. Note that the commit message of 258d2edb claims that it's not out of spec. I _think_ we can do the revert with a good explanation, but I'll leave that to you.

Quoted 8 lines
Quoted 7 lines
Quoted 1 line

(CCed Charles who wrote that commit)

IIUC, the plan is to revert my old commit and use the footer->size field to describe images greater than 127 GB. This change would break other tools from Virtual PC, Xens vhd-util and maybe others from reading images greater than 127 GB because the head field would be forced back to using 16 and these tools won't know to check the footer->size field. Is there any reason not to keep the original commit and still use the footer->size field?

do you have a Pointer to a spec that is newer than 2006?

No, that is the most recent.

Quoted 10 lines

the one i have
describes CHS calculation up to 65535 x 16 x 255 sectors. that is set as Maximum if total sectors is higher.
I would do the same when writing a
footer. in vhd_open I would derive total_sectors from C x H x S except for the case that it is exactly 65535 x 16 x 255. In this case I would take footer->size / 512.
Virtualbox does it that way and at the comment from Stefan in the commit message for your Patch suggest that you observed a similar behaviour for HyperV.

Quoted 3 lines

Right, although this was a long time ago for me to remember the specifics :) In the end I think supporting at least the 2 TB size allowed by the spec is what we need while not breaking existing images.

Thats clear, the absolute limit is 2 TB. We still refuse large images with -EFBIG. Before your patch every image above 127GB was refused with that error.

Would you agree to implement the following:

a) Use footer->size / 512 for bs->total_sectors iff CxHxS === 65535 x 16 x 255 (Kevins RFC Patch).

    Setting these values seems to be the inofficial way to say look at the footer->size.
    Normally I would argue to use footer->size iff CxHxS >== 65535 x 16 x 255, but
    this would break opening of our old out of spec images created with qemu-img.

b) change calculate_geometry() to do exactly what is in the old specs of 2006. In particular this will set CxHxS to 65535 x 16 x 255 for anything >127G and thus tells vpc_open to look at the footer->size.

Cheers,
Peter

#18 lore raw

[meta-comment]

Quoted 2 lines · Peter Lieven
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Am 09.02.2015 um 17:09 schrieb Kevin Wolf:
The CHS calculation as done per the VHD spec imposes a maximum

That's a LOT of nested quoting. It's okay to trim your reply to cut out the portion of the thread that people have already seen 17 times. :)

Quoted 6 lines
Quoted 3 lines

Right, although this was a long time ago for me to remember the specifics :) In the end I think supporting at least the 2 TB size allowed by the spec is what we need while not breaking existing images.

Thats clear, the absolute limit is 2 TB. We still refuse large images with -EFBIG. Before your patch every image above 127GB was refused with that error.

For example, trimming to just this much means that the reply occurs on the first screenful, rather than having to scroll through several pages of quotes to find new content.

Signature · 2 lines
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org