qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough
PATCH v2 1/1 qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough
Fix IGD passthrough logic to properly expose PCH ISA bridge (instead of exposing as pci-pci bridge). The i915 driver require this to correctly detect the PCH version and enable version specific code path.
Timothy Guo <[email protected]>
hw/pci.c +0 -5
@@ -871,11 +871,6 @@ void pci_unplug_netifs(void) } } -typedef struct { -PCIDevice dev; -PCIBus *bus; -}PCIBridge ; - void pci_bridge_write_config(
PCIDevice *d,
uint32_t address, uint32_t val, int len)
{
hw/pci.h +5 -0
@@ -222,6 +222,11 @@ struct PCIDevice {int irq_state[4 ]; }; +typedef struct { +PCIDevice dev; +PCIBus *bus; +}PCIBridge ; +extern char direct_pci_str[];extern int direct_pci_msitranslate;extern int direct_pci_power_mgmt;
hw/pt-graphics.c +4 -3
@@ -3,6 +3,7 @@*/ #include "pass-through.h" +#include "pci.h" #include "pci/header.h" #include "pci/pci.h"
@@ -40,9 +41,26 @@ void intel_pch_init(PCIBus *bus) did =pt_pci_host_read (pci_dev_1f,PCI_DEVICE_ID ,2 ); rid =pt_pci_host_read (pci_dev_1f,PCI_REVISION ,1 ); -if ( vid ==PCI_VENDOR_ID_INTEL ) -pci_bridge_init (bus,PCI_DEVFN (0x1f ,0 ), vid, did, rid, - pch_map_irq,"intel_bridge_1f" ); +if (vid ==PCI_VENDOR_ID_INTEL ) { + PCIBridge *s = (PCIBridge *)pci_register_device(bus, "intel_bridge_1f", + sizeof(PCIBridge), PCI_DEVFN(0x1f, 0), NULL,
pci_bridge_write_config);
+
+ pci_config_set_vendor_id(s->dev.config, vid);
+ pci_config_set_device_id(s->dev.config, did);
+
+ s->dev.config[PCI_COMMAND] = 0x06; // command = bus master, pci mem + s->dev.config[PCI_COMMAND + 1] = 0x00; + s->dev.config[PCI_STATUS] = 0xa0; // status = fast back-to-back, 66MHz, no error + s->dev.config[PCI_STATUS + 1] = 0x00; // status = fast devsel + s->dev.config[PCI_REVISION] = rid; + s->dev.config[PCI_CLASS_PROG] = 0x00; // programming i/f + pci_config_set_class(s->dev.config, PCI_CLASS_BRIDGE_ISA); + s->dev.config[PCI_LATENCY_TIMER] = 0x10; + s->dev.config[PCI_HEADER_TYPE] = 0x80; + s->dev.config[PCI_SEC_STATUS] = 0xa0;
+
+ s->bus = pci_register_secondary_bus(&s->dev, pch_map_irq);
+ }
}
uint32_t igd_read_opregion(struct pt_dev *pci_dev)
Quoted 7 lines · G.R.
Fix IGD passthrough logic to properly expose PCH ISA bridge (instead of exposing as pci-pci bridge). The i915 driver require this to correctly detect the PCH version and enable version specific code path.
Signed-off-by Stefano Stabellini <[email protected]>,Timothy Guo <[email protected]>
The patch is fine by me
Quoted 78 lines
hw/pci.c +0
-5@@ -871,11 +871,6 @@ void pci_unplug_netifs(void) } } -typedef struct { -PCIDevice dev; -PCIBus *bus; -}PCIBridge ; -void pci_bridge_write_config (PCIDevice *d,uint32_t address,uint32_t val,int len){
hw/pci.h +5
-0@@ -222,6 +222,11 @@ struct PCIDevice {int irq_state[4 ]; }; +typedef struct { +PCIDevice dev; +PCIBus *bus; +}PCIBridge ; +extern char direct_pci_str[];extern int direct_pci_msitranslate;extern int direct_pci_power_mgmt;hw/pt-graphics.c +11
-3@@ -3,6 +3,7 @@*/ #include "pass-through.h" +#include "pci.h" #include "pci/header.h" #include "pci/pci.h" @@ -40,9 +41,26 @@ void intel_pch_init(PCIBus *bus) did =pt_pci_host_read (pci_dev_1f,PCI_DEVICE_ID ,2 ); rid =pt_pci_host_read (pci_dev_1f,PCI_REVISION ,1 ); -if ( vid ==PCI_VENDOR_ID_INTEL ) -pci_bridge_init (bus,PCI_DEVFN (0x1f ,0 ), vid, did, rid, - pch_map_irq,"intel_bridge_1f" ); +if (vid ==PCI_VENDOR_ID_INTEL ) { +PCIBridge *s = (PCIBridge *)pci_register_device (bus,"intel_bridge_1f" , +sizeof (PCIBridge),PCI_DEVFN (0x1f ,0 ),NULL , pci_bridge_write_config); + +pci_config_set_vendor_id (s->dev.config, vid); +pci_config_set_device_id (s->dev.config, did); + + s->dev.config[PCI_COMMAND ] =0x06 ;// command = bus master, pci mem + s->dev.config[PCI_COMMAND +1 ] =0x00 ; + s->dev.config[PCI_STATUS ] =0xa0 ;// status = fast back-to-back, 66MHz, no error
hw/pt-graphics.c +10
-0+ s->dev.config[PCI_STATUS +1 ] =0x00 ;// status = fast devsel + s->dev.config[PCI_REVISION ] = rid; + s->dev.config[PCI_CLASS_PROG ] =0x00 ;// programming i/f +pci_config_set_class (s->dev.config,PCI_CLASS_BRIDGE_ISA ); + s->dev.config[PCI_LATENCY_TIMER ] =0x10 ; + s->dev.config[PCI_HEADER_TYPE ] =0x80 ; + s->dev.config[PCI_SEC_STATUS ] =0xa0 ; + + s->bus =pci_register_secondary_bus (&s->dev, pch_map_irq); + } }uint32_t igd_read_opregion (struct pt_dev *pci_dev)
Ian Jackson
AppliedG.R. writes ("[Xen-devel] [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough"):
Quoted 4 lines
Fix IGD passthrough logic to properly expose PCH ISA bridge (instead of exposing as pci-pci bridge). The i915 driver require this to correctly detect the PCH version and enable version specific code path.
Applied, thanks. I fixed up some wrap damage.
Ian.
Quoted 9 lines
Quoted 1 line
Quoted 1 line · Ian Jackson
G.R. writes ("[Xen-devel] [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough"):
Quoted 4 lines
Fix IGD passthrough logic to properly expose PCH ISA bridge (instead of exposing as pci-pci bridge). The i915 driver require this to correctly detect the PCH version and enable version specific code path.
Applied, thanks. I fixed up some wrap damage.
Which means that my pointing out of shortcomings in this patch got completely ignored - we can only hope that this won't bite us later...
Jan
Jan Beulich writes ("Re: [Xen-devel] [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough"):
Quoted 5 lines
Quoted 1 line
Applied, thanks. I fixed up some wrap damage.
Which means that my pointing out of shortcomings in this patch got completely ignored - we can only hope that this won't bite us later...
Sorry about that, perhaps I didn't spot the thread. Do you think it should be reverted ?
Ian.
Quoted 11 lines
Quoted 1 line
Quoted 1 line · Ian Jackson
Jan Beulich writes ("Re: [Xen-devel] [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough"):
Quoted 5 lines
Quoted 1 line
Applied, thanks. I fixed up some wrap damage.
Which means that my pointing out of shortcomings in this patch got completely ignored - we can only hope that this won't bite us later...
Sorry about that, perhaps I didn't spot the thread. Do you think it should be reverted ?
I'd prefer if you did so, and wait for a cleaned up version of the patch to be submitted. If you leave in what's there right now, I would b afraid that we'd never get to see a fixup patch on top.
Jan
Quoted 9 lines · Jan Beulich
Quoted 9 lines
Quoted 1 line
Quoted 1 line · Ian Jackson
Jan Beulich writes ("Re: [Xen-devel] [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough"):
Quoted 5 lines
Quoted 1 line
Applied, thanks. I fixed up some wrap damage.
Which means that my pointing out of shortcomings in this patch got completely ignored - we can only hope that this won't bite us later...
Quoted 6 lines
Quoted 2 lines
Sorry about that, perhaps I didn't spot the thread. Do you think it should be reverted ?
I'd prefer if you did so, and wait for a cleaned up version of the patch to be submitted. If you leave in what's there right now, I would b afraid that we'd never get to see a fixup patch on top.
You are talking about [email protected]? G.R., can you work on this patch a bit more and address Jan's comments? It shouldn't take long..
Re: [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough [and 2 more messages]
Stefano Stabellini writes ("Re: [Xen-devel] [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough"):
Quoted 10 lines · Jan Beulich
Quoted 9 lines
Quoted 8 lines
Quoted 1 line
Quoted 1 line · Ian Jackson
Jan Beulich writes ("Re: [Xen-devel] [PATCH v2] qemu-xen:Correctly expose PCH
Quoted 3 lines
Which means that my pointing out of shortcomings in this patch got completely ignored - we can only hope that this won't bite us later...
Sorry about that, perhaps I didn't spot the thread. Do you think it should be reverted ?
Quoted 7 lines
Quoted 3 lines
I'd prefer if you did so, and wait for a cleaned up version of the patch to be submitted. If you leave in what's there right now, I would b afraid that we'd never get to see a fixup patch on top.
You are talking about [email protected]? G.R., can you work on this patch a bit more and address Jan's comments? It shouldn't take long..
Stefano Stabellini writes ("Re: [Xen-devel] Patch series for IGD passthrough"):
Quoted 15 lines · George Dunlap
Quoted 14 lines · Pasi Kärkkäinen
Quoted 13 lines · Pasi Kärkkäinen
Quoted 12 lines · Rui Guo
Quoted 10 lines
This series contains all the fixes required to produce a working IGD passthrough box. All the changes are previously seen in the dev list but not yet accepted. Some of them are out-dated and need some reshape.
Detailed description can be found later in each patch.
. [PATCH 1/3] qemu-xen-trad/pt_msi_disable: do not clear all MSI flags . [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD . [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Looking at qemu-xen-unstable I think patches 2 and 3 are not yet applied. (patch 2 was applied earlier but it got reverted).
Is it likely that we'll get these merged in for Xen 4.3 ?
Have you addressed Jan's comments from February ? I don't see anything saying you have done so.
Also please could you CC patches to the maintainer (ie, me, in this case) and also to people who have commented on them in the past (Jan, in this case).
Simply constantly pinging is not appropriate.
Ian.
Re: [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough [and 2 more messages]
Quoted 19 lines · Ian Jackson
Stefano Stabellini writes ("Re: [Xen-devel] [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough"):
Quoted 17 lines · Jan Beulich
Quoted 6 lines
Quoted 6 lines
Quoted 1 line
Quoted 1 line · Ian Jackson
Jan Beulich writes ("Re: [Xen-devel] [PATCH v2] qemu-xen:Correctly expose PCH
Quoted 3 lines
Which means that my pointing out of shortcomings in this patch got completely ignored - we can only hope that this won't bite us later...
Quoted 6 lines
Quoted 2 lines
Sorry about that, perhaps I didn't spot the thread. Do you think it should be reverted ?
I'd prefer if you did so, and wait for a cleaned up version of the patch to be submitted. If you leave in what's there right now, I would b afraid that we'd never get to see a fixup patch on top.
You are talking about [email protected]? G.R., can you work on this patch a bit more and address Jan's comments? It shouldn't take long..
So this "PATCH v2" here was posted on 19 Dec 2012. It got applied, and then reverted due to Jan's comments/concerns.
"Patch series for IGD passthrough" has a newer version of this patch, posted on 08 Feb 2013.
Quoted 16 lines
Stefano Stabellini writes ("Re: [Xen-devel] Patch series for IGD passthrough"):
Quoted 15 lines · George Dunlap
Quoted 14 lines · Pasi Kärkkäinen
Quoted 13 lines · Pasi Kärkkäinen
Quoted 12 lines · Rui Guo
Quoted 10 lines
This series contains all the fixes required to produce a working IGD passthrough box. All the changes are previously seen in the dev list but not yet accepted. Some of them are out-dated and need some reshape.
Detailed description can be found later in each patch.
. [PATCH 1/3] qemu-xen-trad/pt_msi_disable: do not clear all MSI flags . [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD . [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific
Quoted 8 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Looking at qemu-xen-unstable I think patches 2 and 3 are not yet applied. (patch 2 was applied earlier but it got reverted).
Is it likely that we'll get these merged in for Xen 4.3 ?
Have you addressed Jan's comments from February ? I don't see anything saying you have done so.
"Patch series for IGD passthrough" has a later version of the patch in question. I'm not sure if Jan's comments are addressed properly, there's no changelog on the patch, but the patch seems different from the one that got applied+reverted.
Quoted 6 lines
Also please could you CC patches to the maintainer (ie, me, in this case) and also to people who have commented on them in the past (Jan, in this case).
Simply constantly pinging is not appropriate.
Sorry for that, but I just wanted to get some attention to these patches, because it wasn't clear what still needs to be done to get them merged.
All three patches on the "Patch series for IGD passthrough" are ACKed by Stefano:
http://lists.xen.org/archives/html/xen-devel/2013-02/msg00593.html
patch 1/3: This one seems to have been merged already to qemu-xen-unstable.
patch 2/3: this one got some comments, but to me it looks like it didn't get a NACK? (and it has Stefano's ACK). patch 3/3: This one might still have un-addressed comments?
Jan: Can you please re-check 2/3 and 3/3 ?
"[PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD passthrough":
http://lists.xen.org/archives/html/xen-devel/2013-02/msg00536.html
"[PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge":
http://lists.xen.org/archives/html/xen-devel/2013-02/msg00538.html
Thanks,
-- Pasi
Re: [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough [and 2 more messages]
Quoted 6 lines
Quoted 1 line
Quoted 1 line · Pasi Kärkkäinen
Jan: Can you please re-check 2/3 and 3/3 ?
"[PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD passthrough":
http://lists.xen.org/archives/html/xen-devel/2013-02/msg00536.html
I'm not sure what you're asking for - my comments there weren't followed up with by the submitter, so I also don't have anything new to add. I don't think I've seen a cleaned up patch so far.
Quoted 3 lines
"[PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge":
http://lists.xen.org/archives/html/xen-devel/2013-02/msg00538.html
Quite similar here, except that Jean was also asked to comment on why some previously existing code is the way it is, and I don't think any explanation was ever given.
I can only second what Ian already said - pinging on a patch that has responses missing from the originator is bad practice. If you wanted to ping anyone, you'd have to ping those who failed to answer questions raised to them, not give the appearance of requesting a patch to be taken that is only half baked.
Jan
_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel
Re: [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough [and 2 more messages]
Quoted 11 lines · Jan Beulich
Quoted 6 lines
Quoted 1 line
Quoted 1 line · Pasi Kärkkäinen
Jan: Can you please re-check 2/3 and 3/3 ?
"[PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD passthrough":
http://lists.xen.org/archives/html/xen-devel/2013-02/msg00536.htmlI'm not sure what you're asking for - my comments there weren't followed up with by the submitter, so I also don't have anything new to add. I don't think I've seen a cleaned up patch so far.
Ok. Thanks for clarifying that (I wasn't sure if there was still unresolved issues or not).
Quoted 8 lines
Quoted 3 lines
"[PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge":
http://lists.xen.org/archives/html/xen-devel/2013-02/msg00538.htmlQuite similar here, except that Jean was also asked to comment on why some previously existing code is the way it is, and I don't think any explanation was ever given.
Ok.
Quoted 6 lines
I can only second what Ian already said - pinging on a patch that has responses missing from the originator is bad practice. If you wanted to ping anyone, you'd have to ping those who failed to answer questions raised to them, not give the appearance of requesting a patch to be taken that is only half baked.
I mostly wanted to clarify the situation with these patches, I'm not suggesting to merge half baked patches.
So thanks for the comments!
-- Pasi
Re: [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough [and 2 more messages]
Quoted 8 lines · Pasi Kärkkäinen
Quoted 6 lines
I can only second what Ian already said - pinging on a patch that has responses missing from the originator is bad practice. If you wanted to ping anyone, you'd have to ping those who failed to answer questions raised to them, not give the appearance of requesting a patch to be taken that is only half baked.
I mostly wanted to clarify the situation with these patches,
Instead of adding more work to our already overburdened maintainers please start in these cases by working with the submitter.
It is the submitters responsibility to respond to queries and to keep pushing the patch as necessary. If the submitter isn't engaged any more then you need to either pick it up yourself or find someone else who is willing to carry the patch forward, there's no point asking the maintainers to clarify the situation if no one is around who is going to act on their responses.
Ian.
_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel
Re: [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough [and 2 more messages]
Quoted 11 lines · Ian Campbell
Quoted 7 lines · Pasi Kärkkäinen
Quoted 6 lines
I can only second what Ian already said - pinging on a patch that has responses missing from the originator is bad practice. If you wanted to ping anyone, you'd have to ping those who failed to answer questions raised to them, not give the appearance of requesting a patch to be taken that is only half baked.
I mostly wanted to clarify the situation with these patches,
Instead of adding more work to our already overburdened maintainers please start in these cases by working with the submitter.
Quoted 6 lines
It is the submitters responsibility to respond to queries and to keep pushing the patch as necessary. If the submitter isn't engaged any more then you need to either pick it up yourself or find someone else who is willing to carry the patch forward, there's no point asking the maintainers to clarify the situation if no one is around who is going to act on their responses.
But if he were going to pick it up, or if he wants to go work with the submitter, who may not have a clear understanding what the delay was about or how the development process works, then he'd first need to figure out what the status was, right?
This sort of "facilitating" role seems like a useful thing to do to allow contributions which would otherwise flounder to make it above the bar into acceptance.
-George
_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel
Re: [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough [and 2 more messages]
Quoted 19 lines · George Dunlap
Quoted 11 lines · Ian Campbell
Quoted 7 lines · Pasi Kärkkäinen
Quoted 6 lines
I can only second what Ian already said - pinging on a patch that has responses missing from the originator is bad practice. If you wanted to ping anyone, you'd have to ping those who failed to answer questions raised to them, not give the appearance of requesting a patch to be taken that is only half baked.
I mostly wanted to clarify the situation with these patches,
Instead of adding more work to our already overburdened maintainers please start in these cases by working with the submitter.
Quoted 6 lines
It is the submitters responsibility to respond to queries and to keep pushing the patch as necessary. If the submitter isn't engaged any more then you need to either pick it up yourself or find someone else who is willing to carry the patch forward, there's no point asking the maintainers to clarify the situation if no one is around who is going to act on their responses.
Quoted 8 lines
But if he were going to pick it up, or if he wants to go work with the submitter, who may not have a clear understanding what the delay was about or how the development process works, then he'd first need to figure out what the status was, right?
This sort of "facilitating" role seems like a useful thing to do to allow contributions which would otherwise flounder to make it above the bar into acceptance.
A facilitator should first ask the submitter what they think is going on, and try and offer them guidance, which could be fairly generic e.g. remind them that the need to resubmit if a patch is dropped, or it could be more specific, i.e. reminding them of the need to respond to particular queries made by the maintainers last time.
Yes, this means they would have needed to have read the background and make an attempt to understand the current status first. They may not be able to figure it out (or might get it wrong) but at least the obvious cases can then be covered.
Facilitating can surely be useful and appreciated. But if "facilitate" just means pinging the maintainers without having spoken to the submitter, read up on the background and followed up on outstanding queries then it is just making more work for maintainers for no gain and isn't really facilitating anything.
Remember, Submitters (and facilitators) scale while maintainers do not.
Ian.
_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel
Re: [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough [and 2 more messages]
Quoted 20 lines · Ian Campbell
Quoted 11 lines
Quoted 10 lines
Quoted 1 line
I mostly wanted to clarify the situation with these patches,
Instead of adding more work to our already overburdened maintainers please start in these cases by working with the submitter.
It is the submitters responsibility to respond to queries and to keep pushing the patch as necessary. If the submitter isn't engaged any more then you need to either pick it up yourself or find someone else who is willing to carry the patch forward, there's no point asking the maintainers to clarify the situation if no one is around who is going to act on their responses.
Quoted 8 lines
But if he were going to pick it up, or if he wants to go work with the submitter, who may not have a clear understanding what the delay was about or how the development process works, then he'd first need to figure out what the status was, right?
This sort of "facilitating" role seems like a useful thing to do to allow contributions which would otherwise flounder to make it above the bar into acceptance.
Quoted 6 lines
A facilitator should first ask the submitter what they think is going on, and try and offer them guidance, which could be fairly generic e.g. remind them that the need to resubmit if a patch is dropped, or it could be more specific, i.e. reminding them of the need to respond to particular queries made by the maintainers last time.
Yep, I should have talked to G.R. first. Sorry about that.
Quoted 5 lines
Yes, this means they would have needed to have read the background and make an attempt to understand the current status first. They may not be able to figure it out (or might get it wrong) but at least the obvious cases can then be covered.
I did read all the emails in various threads related to these patches, and I tried understanding where we are.
It was a bit confusing because for example one of the patches got applied, then reverted, re-submitted after that, and I didn't 100% understand from the comments if there are still unresolved issues or not.
Now it's clear that not all of the issues are sorted out yet.
(Btw right now we're discussing in a thread that doesn't have the latest version of the patches..)
Quoted 8 lines
Facilitating can surely be useful and appreciated. But if "facilitate" just means pinging the maintainers without having spoken to the submitter, read up on the background and followed up on outstanding queries then it is just making more work for maintainers for no gain and isn't really facilitating anything.
Remember, Submitters (and facilitators) scale while maintainers do not.
I do understand that. Thanks for your patience :)
-- Pasi
Re: [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough [and 2 more messages]
In reply to George DunlapQuoted 5 lines
Quoted 1 line
Quoted 1 line · George Dunlap
But if he were going to pick it up, or if he wants to go work with the submitter, who may not have a clear understanding what the delay was about or how the development process works, then he'd first need to figure out what the status was, right?
But that's what is available in the list archives - in the case at hand, Pasi pointed at patches that had trees of subsequent communication attached to them, and in all cases the leaves where unanswered questions. So figuring out the status was quite strait forward here.
I admit that this may not always be as simple (or otherwise I likely wouldn't even have bothered looking at the old conversation again).
Bottom line is that I think we should be allowed to expect homework to be done before pinging maintainers.
Jan
Jan Beulich writes ("Re: [Xen-devel] [PATCH v2] qemu-xen:Correctly expose PCH ISA bridge for IGD passthrough"):
Quoted 7 lines
[Ian Jackson:]
Quoted 2 lines
Sorry about that, perhaps I didn't spot the thread. Do you think it should be reverted ?
I'd prefer if you did so, and wait for a cleaned up version of the patch to be submitted. If you leave in what's there right now, I would b afraid that we'd never get to see a fixup patch on top.
Done.
Ian.
Patch series for IGD passthrough
This series contains all the fixes required to produce a working IGD passthrough box. All the changes are previously seen in the dev list but not yet accepted. Some of them are out-dated and need some reshape.
Detailed description can be found later in each patch.
. [PATCH 1/3] qemu-xen-trad/pt_msi_disable: do not clear all MSI flags . [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD . [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific
Thanks,
Timothy
PATCH 1/3 qemu-xen-trad/pt_msi_disable: do not clear all MSI flags
"qemu-xen-trad: fix msi_translate with PV event delivery" added a pt_msi_disable() call into pt_msgctrl_reg_write, clearing the MSI flags as a consequence. MSIs get enabled again soon after by calling pt_msi_setup.
However the MSI flags are only setup once in the pt_msgctrl_reg_init function, so from the QEMU point of view the device has lost some important properties, like for example PCI_MSI_FLAGS_64BIT.
This patch fixes the bug by clearing only the MSI enabled/mapped/initialized flags in pt_msi_disable.
Xen-devel: http://marc.info/?l=xen-devel&m=135489879503075
--- hw/pt-msi.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
hw/pt-msi.c +1 -1
@@ -213,7 +213,7 @@ void pt_msi_disable(struct pt_dev *dev) out:/* clear msi info */ - dev->msi->flags =0 ; + dev->msi->flags &= ~(MSI_FLAG_UNINIT |PT_MSI_MAPPED |PCI_MSI_FLAGS_ENABLE ); dev->msi->pirq =-1 ; dev->msi_trans_en =0 ; }
Signature · 1 line
1.7.10.4
PATCH 2/3 qemu-xen-trad: Correctly expose PCH ISA bridge for IGD passthrough
In reply to Rui GuoThe i915 driver probes chip version through PCH ISA bridge device / vendor ID. Previously, the PCH ISA bridge is exposed as PCI-PCI bridge in qemu-xen-trad, which breaks the assumption of the driver. This change fixes the issue by correctly exposing the ISA bridge to domU.
Rui Guo <[email protected]>
Xen-devel: http://marc.info/?l=xen-devel&m=135548433715750
--- hw/pci.c | 5 ----- hw/pci.h | 5 +++++ hw/pt-graphics.c | 24 +++++++++++++++++++++--- 3 files changed, 26 insertions(+), 8 deletions(-)
hw/pci.c +0 -5
@@ -871,11 +871,6 @@ void pci_unplug_netifs(void) } } -typedef struct { -PCIDevice dev; -PCIBus *bus; -}PCIBridge ; -void pci_bridge_write_config (PCIDevice *d,uint32_t address,uint32_t val,int len) {
hw/pci.h +5 -0
@@ -222,6 +222,11 @@ struct PCIDevice {int irq_state[4 ]; }; +typedef struct { +PCIDevice dev; +PCIBus *bus; +}PCIBridge ; +extern char direct_pci_str[];extern int direct_pci_msitranslate;extern int direct_pci_power_mgmt;
hw/pt-graphics.c +21 -3
@@ -3,6 +3,7 @@*/ #include "pass-through.h" +#include "pci.h" #include "pci/header.h" #include "pci/pci.h"
@@ -40,9 +41,26 @@ void intel_pch_init(PCIBus *bus) did =pt_pci_host_read (pci_dev_1f,PCI_DEVICE_ID ,2 ); rid =pt_pci_host_read (pci_dev_1f,PCI_REVISION ,1 ); -if ( vid ==PCI_VENDOR_ID_INTEL ) -pci_bridge_init (bus,PCI_DEVFN (0x1f ,0 ), vid, did, rid, - pch_map_irq,"intel_bridge_1f" ); +if (vid ==PCI_VENDOR_ID_INTEL ) { +PCIBridge *s = (PCIBridge *)pci_register_device (bus,"intel_bridge_1f" , +sizeof (PCIBridge),PCI_DEVFN (0x1f ,0 ),NULL , pci_bridge_write_config); + +pci_config_set_vendor_id (s->dev.config, vid); +pci_config_set_device_id (s->dev.config, did); + + s->dev.config[PCI_COMMAND ] =0x06 ;// command = bus master, pci mem + s->dev.config[PCI_COMMAND +1 ] =0x00 ; + s->dev.config[PCI_STATUS ] =0xa0 ;// status = fast back-to-back, 66MHz, no error + s->dev.config[PCI_STATUS +1 ] =0x00 ;// status = fast devsel + s->dev.config[PCI_REVISION ] = rid; + s->dev.config[PCI_CLASS_PROG ] =0x00 ;// programming i/f +pci_config_set_class (s->dev.config,PCI_CLASS_BRIDGE_ISA ); + s->dev.config[PCI_LATENCY_TIMER ] =0x10 ; + s->dev.config[PCI_HEADER_TYPE ] =0x80 ; + s->dev.config[PCI_SEC_STATUS ] =0xa0 ; + + s->bus =pci_register_secondary_bus (&s->dev, pch_map_irq); + } }uint32_t igd_read_opregion (struct pt_dev *pci_dev)
Signature · 1 line
1.7.10.4
Re: [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD passthrough
Quoted 21 lines
Quoted 1 line
Quoted 1 line · Rui Guo
hw/pt-graphics.c +13
-3@@ -40,9 +41,26 @@ void intel_pch_init(PCIBus *bus) did =pt_pci_host_read (pci_dev_1f,PCI_DEVICE_ID ,2 ); rid =pt_pci_host_read (pci_dev_1f,PCI_REVISION ,1 ); -if ( vid ==PCI_VENDOR_ID_INTEL ) -pci_bridge_init (bus,PCI_DEVFN (0x1f ,0 ), vid, did, rid, - pch_map_irq,"intel_bridge_1f" ); +if (vid ==PCI_VENDOR_ID_INTEL ) { +PCIBridge *s = (PCIBridge *)pci_register_device (bus,"intel_bridge_1f" , +sizeof (PCIBridge),PCI_DEVFN (0x1f ,0 ),NULL , pci_bridge_write_config); + +pci_config_set_vendor_id (s->dev.config, vid); +pci_config_set_device_id (s->dev.config, did); + + s->dev.config[PCI_COMMAND ] =0x06 ;// command = bus master, pci mem + s->dev.config[PCI_COMMAND +1 ] =0x00 ; + s->dev.config[PCI_STATUS ] =0xa0 ;// status = fast back-to-back, 66MHz, no error + s->dev.config[PCI_STATUS +1 ] =0x00 ;// status = fast devsel + s->dev.config[PCI_REVISION ] = rid; + s->dev.config[PCI_CLASS_PROG ] =0x00 ;// programming i/f
hw/pt-graphics.c void intel_pch_init(PCIBus *bus) · line 66
Quoted 10 lines
hw/pt-graphics.c +7
-0+pci_config_set_class (s->dev.config,PCI_CLASS_BRIDGE_ISA ); + s->dev.config[PCI_LATENCY_TIMER ] =0x10 ; + s->dev.config[PCI_HEADER_TYPE ] =0x80 ; + s->dev.config[PCI_SEC_STATUS ] =0xa0 ; + + s->bus =pci_register_secondary_bus (&s->dev, pch_map_irq); + } }uint32_t igd_read_opregion (struct pt_dev *pci_dev)
For one I wonder whether this is really _always_ correct. I.e. the device at 00:1f.0 always being an ISA bridge. Wouldn't it be better to mirror whatever the host device says?
And then I don't see why you need to open code all of pci_bridge_init() instead of just overriding the class value after you called that function (or, if the order of events for some reason matters, adding another parameter to the function).
Jan
Re: [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD passthrough
Quoted 6 lines
For one I wonder whether this is really _always_ correct. I.e. the device at 00:1f.0 always being an ISA bridge. Wouldn't it be better to mirror whatever the host device says?
From the Intel driver code, it's looking for an intel ISA bridge.
So your question would be should it be always at 00:1f.0. So far it seems that it is.
Quoted 4 lines
And then I don't see why you need to open code all of
pci_bridge_init()instead of just overriding the class value after you called that function (or, if the order of events for some reason matters, adding another parameter to the function).
Stefano, could you comment? It's your code after all.
Thanks,
Timothy (Rui)
Re: [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD passthrough
Quoted 7 lines
Quoted 6 lines
Quoted 1 line · G.R.
For one I wonder whether this is really _always_ correct. I.e. the device at 00:1f.0 always being an ISA bridge. Wouldn't it be better to mirror whatever the host device says?
From the Intel driver code, it's looking for an intel ISA bridge.
That doesn't mean that it always will be.
Quoted 2 lines
So your question would be should it be always at 00:1f.0. So far it seems that it is.
Same thing here. We ought to be careful, or else we risk to introduce issues that pretty hard to locate, debug, and fix.
Jan
Re: [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD passthrough
It has been a long time, but Pasi reminded me to follow this up. Here is my feedback to your concern:
Quoted 9 lines · Jan Beulich
Quoted 7 lines
Quoted 6 lines
Quoted 1 line · G.R.
For one I wonder whether this is really _always_ correct. I.e. the device at 00:1f.0 always being an ISA bridge. Wouldn't it be better to mirror whatever the host device says?
From the Intel driver code, it's looking for an intel ISA bridge.
That doesn't mean that it always will be.
Unless you can 100% simulate the HW, you have to rely on the known protocol between driver && HW.
I agree that they may switch to different protocol some day, but I don't think we have any better choice.
Quoted 6 lines
Quoted 3 lines
So your question would be should it be always at 00:1f.0. So far it seems that it is.
Same thing here. We ought to be careful, or else we risk to introduce issues that pretty hard to locate, debug, and fix.
Since most (if not all) recent intel chipsets in the market have ISA bridge at address 00:1f.0, simulate one at the same address to the guest won't be so bad, given the current protocol between driver && HW.
But I guess your concern is about the hard-coded '00:1f.0' address. Yes, I agree that this is not beautiful at all.
I don't mind changing it to probe the ISA bridge from host. But I'm not familiar with qemu at all, could you show me the API to achieve this purpose?
Also, if we really care about doing the 'correct' thing, I think we should get rid of the default ISA bridge provided by qemu -- currently it requires extra patches to linux i915 driver to work around. Anyway to achieve this purpose?
Thanks,
Timothy (Rui)
Re: [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD passthrough
Quoted 24 lines · G.R.
It has been a long time, but Pasi reminded me to follow this up. Here is my feedback to your concern:
Quoted 9 lines · Jan Beulich
Quoted 7 lines
Quoted 6 lines
Quoted 1 line · G.R.
For one I wonder whether this is really _always_ correct. I.e. the device at 00:1f.0 always being an ISA bridge. Wouldn't it be better to mirror whatever the host device says?
From the Intel driver code, it's looking for an intel ISA bridge.
That doesn't mean that it always will be.
Unless you can 100% simulate the HW, you have to rely on the known protocol between driver && HW.
I agree that they may switch to different protocol some day, but I don't think we have any better choice.Quoted 6 lines
Quoted 3 lines
So your question would be should it be always at 00:1f.0. So far it seems that it is.
Same thing here. We ought to be careful, or else we risk to introduce issues that pretty hard to locate, debug, and fix.
Quoted 9 lines
Since most (if not all) recent intel chipsets in the market have ISA bridge at address 00:1f.0, simulate one at the same address to the guest won't be so bad, given the current protocol between driver && HW.
But I guess your concern is about the hard-coded '00:1f.0' address. Yes, I agree that this is not beautiful at all.
I don't mind changing it to probe the ISA bridge from host. But I'm not familiar with qemu at all, could you show me the API to achieve this purpose?
Hi,
if I got no response on the API that I queried for, I would just send out a patch version that does not touch this part.
Quoted 4 lines
Also, if we really care about doing the 'correct' thing, I think we should get rid of the default ISA bridge provided by qemu -- currently it requires extra patches to linux i915 driver to work around. Anyway to achieve this purpose?
I found one API pci_hide_device() may be helpful for this purpose. I would like to try out, but still lack of the API to locate specified device in the bus.
Quoted 3 lines
Thanks,
Timothy (Rui)
Re: [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD passthrough
In reply to G.R.Quoted 9 lines · G.R.
Quoted 6 lines
For one I wonder whether this is really _always_ correct. I.e. the device at 00:1f.0 always being an ISA bridge. Wouldn't it be better to mirror whatever the host device says?
From the Intel driver code, it's looking for an intel ISA bridge.
So your question would be should it be always at 00:1f.0. So far it seems that it is.
Quoted 6 lines
Quoted 4 lines
And then I don't see why you need to open code all of
pci_bridge_init()instead of just overriding the class value after you called that function (or, if the order of events for some reason matters, adding another parameter to the function).Stefano, could you comment? It's your code after all.
It's not trivial because PCIBus is defined in hw/pci.c, so you can't dereference members of PCIBus from hw/pt-graphics.c.
However if you come up with a good patch that makes it work, even better.
Re: [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD passthrough
Quoted 1 line
Quoted 1 line
Quoted 1 line
On 08.02.13 at 12:30, Stefano Stabellini <[email protected]>
Quoted 15 lines · wrote:
Quoted 14 lines · G.R.
Quoted 6 lines
For one I wonder whether this is really _always_ correct. I.e. the device at 00:1f.0 always being an ISA bridge. Wouldn't it be better to mirror whatever the host device says?
From the Intel driver code, it's looking for an intel ISA bridge.
So your question would be should it be always at 00:1f.0. So far it seems that it is.
Quoted 4 lines
And then I don't see why you need to open code all of
pci_bridge_init()instead of just overriding the class value after you called that function (or, if the order of events for some reason matters, adding another parameter to the function).
Quoted 4 lines
Quoted 1 line
Stefano, could you comment? It's your code after all.
It's not trivial because PCIBus is defined in
hw/pci.c, so you can't dereference members of PCIBus fromhw/pt-graphics.c.
But the patch makes it so the structure itself becomes public? What problem do you see?
Jan
Re: [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD passthrough
Quoted 17 lines · Jan Beulich
Quoted 1 line
Quoted 1 line
Quoted 1 line
On 08.02.13 at 12:30, Stefano Stabellini <[email protected]>
Quoted 15 lines · wrote:
Quoted 14 lines · G.R.
Quoted 6 lines
For one I wonder whether this is really _always_ correct. I.e. the device at 00:1f.0 always being an ISA bridge. Wouldn't it be better to mirror whatever the host device says?
From the Intel driver code, it's looking for an intel ISA bridge.
So your question would be should it be always at 00:1f.0. So far it seems that it is.
Quoted 4 lines
And then I don't see why you need to open code all of
pci_bridge_init()instead of just overriding the class value after you called that function (or, if the order of events for some reason matters, adding another parameter to the function).
Quoted 7 lines
Quoted 4 lines
Quoted 1 line
Stefano, could you comment? It's your code after all.
It's not trivial because PCIBus is defined in
hw/pci.c, so you can't dereference members of PCIBus fromhw/pt-graphics.c.But the patch makes it so the structure itself becomes public? What problem do you see?
QEMU's include chain can be devious at times but I have no objections in making PCIBus public.
Re: [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD passthrough
Quoted 20 lines · Stefano Stabellini
Quoted 19 lines · Jan Beulich
Quoted 1 line
Quoted 1 line
Quoted 1 line
On 08.02.13 at 12:30, Stefano Stabellini <[email protected]>
Quoted 17 lines · wrote:
Quoted 15 lines · G.R.
Quoted 6 lines
For one I wonder whether this is really _always_ correct. I.e. the device at 00:1f.0 always being an ISA bridge. Wouldn't it be better to mirror whatever the host device says?
From the Intel driver code, it's looking for an intel ISA bridge.
So your question would be should it be always at 00:1f.0. So far it seems that it is.
Quoted 4 lines
And then I don't see why you need to open code all of
pci_bridge_init()instead of just overriding the class value after you called that function (or, if the order of events for some reason matters, adding another parameter to the function).Stefano, could you comment? It's your code after all.
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines
It's not trivial because PCIBus is defined in
hw/pci.c, so you can't dereference members of PCIBus fromhw/pt-graphics.c.But the patch makes it so the structure itself becomes public? What problem do you see?
QEMU's include chain can be devious at times but I have no objections in making PCIBus public.
Are you talking about making PCIBus public (to hw/pci.h) and change the code to something like this:
if (vid == PCI_VENDOR_ID_INTEL) {
PCIBus *s = pci_bridge_init(bus, PCI_DEVFN(0x1f, 0), vid, did, rid,
pch_map_irq, "intel_bridge_1f");
pci_config_set_class(s->parent_dev.config, PCI_CLASS_BRIDGE_ISA);
s->parent_dev.config[PCI_HEADER_TYPE] = 0x80;
}Re: [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD passthrough
Quoted 35 lines
Quoted 1 line
Quoted 1 line · G.R.
Quoted 20 lines · Stefano Stabellini
Quoted 19 lines · Jan Beulich
Quoted 1 line
Quoted 1 line
Quoted 1 line
On 08.02.13 at 12:30, Stefano Stabellini <[email protected]>
Quoted 17 lines · wrote:
Quoted 15 lines · G.R.
Quoted 6 lines
For one I wonder whether this is really _always_ correct. I.e. the device at 00:1f.0 always being an ISA bridge. Wouldn't it be better to mirror whatever the host device says?
From the Intel driver code, it's looking for an intel ISA bridge.
So your question would be should it be always at 00:1f.0. So far it seems that it is.
Quoted 4 lines
And then I don't see why you need to open code all of
pci_bridge_init()instead of just overriding the class value after you called that function (or, if the order of events for some reason matters, adding another parameter to the function).Stefano, could you comment? It's your code after all.
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines
It's not trivial because PCIBus is defined in
hw/pci.c, so you can't dereference members of PCIBus fromhw/pt-graphics.c.But the patch makes it so the structure itself becomes public? What problem do you see?
QEMU's include chain can be devious at times but I have no objections in making PCIBus public.
Are you talking about making PCIBus public (to
hw/pci.h) and change the code to something like this:
Quoted 7 lines
if (vid == PCI_VENDOR_ID_INTEL) {
PCIBus *s = pci_bridge_init(bus, PCI_DEVFN(0x1f, 0), vid, did, rid,
pch_map_irq, "intel_bridge_1f");
pci_config_set_class(s->parent_dev.config, PCI_CLASS_BRIDGE_ISA);
s->parent_dev.config[PCI_HEADER_TYPE] = 0x80;
}Along those lines, yes.
Jan
PATCH 3/3 qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
In reply to Rui GuoSome versions of the Windows Intel GPU driver expect the vendor PCI capability to be there on the host bridge config space when passing through a Intel GPU.
As part of the change, the init for pt_pci_host() return value has to be modified. With an init of -1 all the return value smaller than a double word will be prefixed with "f"s.
Rui Guo <[email protected]>
Xen-devel: http://marc.info/?l=xen-devel&m=135748187808766
--- hw/pass-through.c | 2 +- hw/pt-graphics.c | 69 ++++++++++++++++++++++++++++++++++++++++++++++++----- 2 files changed, 64 insertions(+), 7 deletions(-)
hw/pass-through.c +1 -1
@@ -2101,7 +2101,7 @@ struct pci_dev *pt_pci_get_dev(int bus, int dev, int fn)u32 pt_pci_host_read (struct pci_dev *pci_dev,u32 addr,int len) { -u32 val =-1 ; +u32 val =0 ;pci_access_init ();pci_read_block (pci_dev, addr, (u8 *) &val, len);
hw/pt-graphics.c +63 -6
103 lines not on this page · raw message
Signature · 1 line
1.7.10.4
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 19 lines
Quoted 1 line
Quoted 1 line · Rui Guo
hw/pt-graphics.c +18
-0+static uint32_t igd_pci_read_vendor_cap (PCIDevice *pci_dev,uint32_t config_addr,int len, +uint32_t *val) +{ +struct pci_dev *pci_dev_host_bridge =pt_pci_get_dev (0 ,0 ,0 ); +uint32_t vendor_cap =0 ; +uint32_t cap_type =0 ; +uint32_t cap_size =0 ; + + vendor_cap =pt_pci_host_read (pci_dev_host_bridge,PCI_INTEL_VENDOR_CAP ,1 ); +if (!vendor_cap) +return 0 ; + + cap_type =pt_pci_host_read (pci_dev_host_bridge, vendor_cap,1 ); +if (cap_type !=PCI_INTEL_VENDOR_CAP_TYPE ) +return 0 ; + +if (config_addr ==PCI_INTEL_VENDOR_CAP ) + {
hw/pt-graphics.c write_default: · line 181
Quoted 10 lines
hw/pt-graphics.c +10
-0+ *val = vendor_cap; +return 1 ; + } + +/* Remove the next capability link */ +if (config_addr == vendor_cap +1 ) + { + *val =0 ; +return 1 ; + }
It doesn't look right to ignore "len" up to here? What if this is a word read at "vendor_cap"?
Also, why would removing the next capability be correct here, when you're not removing _all_ other capabilities?
hw/pt-graphics.c write_default: · line 189
Quoted 8 lines
hw/pt-graphics.c +8
-0+ + cap_size =pt_pci_host_read (pci_dev_host_bridge, vendor_cap +2 ,1 ); +if (config_addr >= vendor_cap && + config_addr + len <= vendor_cap + cap_size) + { + *val =pt_pci_host_read (pci_dev_host_bridge, config_addr, len); +return 1 ; + }
Similarly such a read wouldn't get handled consistently (with the above intention) here: You're not masking off the byte at "vendor_cap + 1".
hw/pt-graphics.c uint32_t igd_pci_read(PCIDevice *pci_dev, uint32_t config_addr, int len) · line 210
Quoted 5 lines
hw/pt-graphics.c +5
-0+/* Exposing writable register does not lead to security risk since this +only apply to read. This may confuse the guest, but it works good so far. +Will switch to mask & merge style only if this is proved broken. +Note: Always expose aligned address if any byte of the dword is to be +exposed. This provides a consistent view, at least for read. */
Such a comment is certainly not helping being confident in the changes you're trying to get merged in.
Jan
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 13 lines
hw/pt-graphics.c write_default: · line 181
Quoted 12 lines
hw/pt-graphics.c +12
-0+if (config_addr ==PCI_INTEL_VENDOR_CAP ) + { + *val = vendor_cap; +return 1 ; + } + +/* Remove the next capability link */ +if (config_addr == vendor_cap +1 ) + { + *val =0 ; +return 1 ; + }
Quoted 2 lines
It doesn't look right to ignore "len" up to here? What if this is a word read at "
vendor_cap"?
The function is for intel IGD only and the first two returns are only sanity checks.
The third is fine also since the PCI register is only one byte in length and other bytes in that word are reserved.
Quoted 3 lines
Also, why would removing the next capability be correct here, when you're not removing _all_ other capabilities?
I need help from the original author.
Jean, could you comment on this line in your original patch?
Quoted 10 lines
hw/pt-graphics.c write_default: · line 189
Quoted 9 lines
hw/pt-graphics.c +8
-0+ + cap_size =pt_pci_host_read (pci_dev_host_bridge, vendor_cap +2 ,1 ); +if (config_addr >= vendor_cap && + config_addr + len <= vendor_cap + cap_size) + { + *val =pt_pci_host_read (pci_dev_host_bridge, config_addr, len); +return 1 ; + }
Quoted 4 lines
Similarly such a read wouldn't get handled consistently (with the above intention) here: You're not masking off the byte at "
vendor_cap+ 1".
Same as above. Jean, could you comment?
Quoted 8 lines
hw/pt-graphics.c uint32_t igd_pci_read(PCIDevice *pci_dev, uint32_t config_addr, int len) · line 210
Quoted 5 lines
hw/pt-graphics.c +5
-0+/* Exposing writable register does not lead to security risk since this +only apply to read. This may confuse the guest, but it works good so far. +Will switch to mask & merge style only if this is proved broken. +Note: Always expose aligned address if any byte of the dword is to be +exposed. This provides a consistent view, at least for read. */ Such a comment is certainly not helping being confident in the changes you're trying to get merged in.
Could you be more specific on your suggestion?
I don't think you are recommending to remove them.
Some clarification: The 'Note' sentence just make it explicit the rule that existing code follows.
And the first sentence about the writeable register are for the 'command' register, which is a side effect of the same rule due to the 'status' register.
Thanks,
Timothy (Rui)
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 14 lines
Quoted 14 lines
Quoted 13 lines · G.R.
hw/pt-graphics.c +12
-0+if (config_addr ==PCI_INTEL_VENDOR_CAP ) + { + *val = vendor_cap; +return 1 ; + } + +/* Remove the next capability link */ +if (config_addr == vendor_cap +1 ) + { + *val =0 ; +return 1 ; + }
Quoted 6 lines
Quoted 2 lines
It doesn't look right to ignore "len" up to here? What if this is a word read at "
vendor_cap"?The function is for intel IGD only and the first two returns are only sanity checks.
The third is fine also since the PCI register is only one byte in length and other bytes in that word are reserved.
Just as for the other patch - this is way too lax an approach. You mustn't introduce dependencies on current or observed behavior, as this may break with future chipsets. You need to follow what actual hardware does (i.e. allow multi-byte aligned reads to properly return _all_ covered registers' values).
Quoted 5 lines
Quoted 2 lines
Review
Quoted 2 lines
(quoted diff) +2
-0+ /* Exposing writable register does not lead to security risk since this + only apply to read. This may confuse the guest, but it works good sofar.
Quoted 2 lines
Review
Quoted 2 lines
(quoted diff) +2
-0+ Will switch to mask & merge style only if this is proved broken. + Note: Always expose aligned address if any byte of the dword is to
Quoted 8 lines
be
Quoted 4 lines
Quoted 1 line
+ exposed. This provides a consistent view, at least for read. */
Such a comment is certainly not helping being confident in the changes you're trying to get merged in.
Could you be more specific on your suggestion? I don't think you are recommending to remove them.
Certainly not. I'm recommending to fix the code to make the comment unnecessary.
Jan
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
I'm going to rework this patch to address Jan's concern. Here is my proposal, please review and comment before I begin:
The proposal is to read a shadow copy of the exposed host register into the config space of the emulated host bridge and relies on the pci_default_read_config() function
to provide proper access.
This methodology only works for constant values, which is our case here. The exposed value is either read-only or write-locked (for BIOS).
The only exception is that the PAVPC (0x58) register is write-locked but not for BIOS.
This is exposed for RW and my proposal is to perform write-through in the register write-support.
Quoted 3 lines
Also, why would removing the next capability be correct here, when you're not removing _all_ other capabilities?
I have no answer about this question. *Jean*, could you help comment since this is from your code?
Thanks,
Timothy
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 8 lines · G.R.
I'm going to rework this patch to address Jan's concern. Here is my proposal, please review and comment before I begin:
The proposal is to read a shadow copy of the exposed host register into the config space of the emulated host bridge and relies on the
pci_default_read_config()function
to provide proper access.
Quoted 5 lines
This methodology only works for constant values, which is our case here. The exposed value is either read-only or write-locked (for BIOS).
The only exception is that the PAVPC (
0x58) register is write-locked but not for BIOS.
So only SeaBIOS or hvmloader should touch it?
Quoted 2 lines
This is exposed for RW and my proposal is to perform write-through in the register write-support.
What does PAVPC do? As in if the driver wrote 0xdeadbeef in there what would happen? Is there a list of the appropiate values it should accept?
Quoted 6 lines
Quoted 4 lines
Also, why would removing the next capability be correct here, when you're not removing _all_ other capabilities?
I have no answer about this question. *Jean*, could you help comment since this is from your code?
If he doesn't answer - if you don't remove the capability does it still work?
Quoted 9 lines
Thanks,
Timothy_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 26 lines · Konrad Rzeszutek Wilk
Quoted 8 lines · G.R.
I'm going to rework this patch to address Jan's concern. Here is my proposal, please review and comment before I begin:
The proposal is to read a shadow copy of the exposed host register into the config space of the emulated host bridge and relies on the
pci_default_read_config()function
to provide proper access.Quoted 5 lines
This methodology only works for constant values, which is our case here. The exposed value is either read-only or write-locked (for BIOS).
The only exception is that the PAVPC (
0x58) register is write-locked but not for BIOS.So only SeaBIOS or hvmloader should touch it?
Quoted 2 lines
This is exposed for RW and my proposal is to perform write-through in the register write-support.
What does PAVPC do? As in if the driver wrote
0xdeadbeefin there what would happen? Is there a list of the appropiate values it should accept?Quoted 2 lines
Quoted 2 lines
Quoted 8 lines
Quoted 4 lines
Quoted 2 lines
Also, why would removing the next capability be correct here, when you're not removing _all_ other capabilities?
I have no answer about this question. *Jean*, could you help comment since this is from your code?
If he doesn't answer - if you don't remove the capability does it still work?
So I actually originally found this issue with the vendor capabilities and created the original patch for it. This was quite some time ago so I had to go back and look. IIRC the vendor specific capabilities were always the first one in the chain and the unchaining code would drop all further capabilities (which we did not want to pass directly to the guest).
We never saw a configuration where the vendor capabilities were not the first. I guess the suggestion is that to make the patch consistent, preceding capabilities should be detected and handled. I am not sure what the best way to do it would be. Perhaps scanning through the chain until 0x09 is found and reporting its offset through 0x34 instead of what is there would be the way to go. Then unchain anything past the 0x09 caps too as is currently done.
Quoted 14 lines
Quoted 9 lines
Thanks,
Timothy_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 36 lines · Ross Philipson
Quoted 26 lines · Konrad Rzeszutek Wilk
Quoted 8 lines · G.R.
I'm going to rework this patch to address Jan's concern. Here is my proposal, please review and comment before I begin:
The proposal is to read a shadow copy of the exposed host register into the config space of the emulated host bridge and relies on the
pci_default_read_config()function
to provide proper access.Quoted 5 lines
This methodology only works for constant values, which is our case here. The exposed value is either read-only or write-locked (for BIOS).
The only exception is that the PAVPC (
0x58) register is write-locked but not for BIOS.So only SeaBIOS or hvmloader should touch it?
Quoted 2 lines
This is exposed for RW and my proposal is to perform write-through in the register write-support.
What does PAVPC do? As in if the driver wrote
0xdeadbeefin there what would happen? Is there a list of the appropiate values it should accept?Quoted 2 lines
Quoted 2 lines
Quoted 8 lines
Quoted 4 lines
Quoted 2 lines
Also, why would removing the next capability be correct here, when you're not removing _all_ other capabilities?
I have no answer about this question. *Jean*, could you help comment since this is from your code?
If he doesn't answer - if you don't remove the capability does it still work?
Quoted 6 lines
So I actually originally found this issue with the vendor capabilities and created the original patch for it. This was quite some time ago so I had to go back and look. IIRC the vendor specific capabilities were always the first one in the chain and the unchaining code would drop all further capabilities (which we did not want to pass directly to the guest).
OK, so blacklisting.
Quoted 8 lines
We never saw a configuration where the vendor capabilities were not the first. I guess the suggestion is that to make the patch consistent, preceding capabilities should be detected and handled. I am not sure what the best way to do it would be. Perhaps scanning through the chain until
0x09is found and reporting its offset through0x34instead of what is there would be the way to go. Then unchain anything past the0x09caps too as is currently done.
Or just scan through the capabilities, and chain only the ones that we want to "Whitelist" and the rest are to be blacklisted. The rest can also have its values set to some bogus value (0xdeadbeef?) Perhaps only when built with 'debug=y'.
Anyhow, if we retain the location (aka offset) of the capabilities then if there is some silly code that expects those to be at specific locations it should still work?
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 54 lines · Konrad Rzeszutek Wilk
Quoted 36 lines · Ross Philipson
Quoted 26 lines · Konrad Rzeszutek Wilk
Quoted 8 lines · G.R.
I'm going to rework this patch to address Jan's concern. Here is my proposal, please review and comment before I begin:
The proposal is to read a shadow copy of the exposed host register into the config space of the emulated host bridge and relies on the
pci_default_read_config()function
to provide proper access.Quoted 5 lines
This methodology only works for constant values, which is our case here. The exposed value is either read-only or write-locked (for BIOS).
The only exception is that the PAVPC (
0x58) register is write-locked but not for BIOS.So only SeaBIOS or hvmloader should touch it?
Quoted 2 lines
This is exposed for RW and my proposal is to perform write-through in the register write-support.
What does PAVPC do? As in if the driver wrote
0xdeadbeefin there what would happen? Is there a list of the appropiate values it should accept?Quoted 2 lines
Quoted 2 lines
Quoted 8 lines
Quoted 4 lines
Quoted 2 lines
Also, why would removing the next capability be correct here, when you're not removing _all_ other capabilities?
I have no answer about this question. *Jean*, could you help comment since this is from your code?
If he doesn't answer - if you don't remove the capability does it still work?
Quoted 6 lines
So I actually originally found this issue with the vendor capabilities and created the original patch for it. This was quite some time ago so I had to go back and look. IIRC the vendor specific capabilities were always the first one in the chain and the unchaining code would drop all further capabilities (which we did not want to pass directly to the guest).
OK, so blacklisting.
Quoted 8 lines
We never saw a configuration where the vendor capabilities were not the first. I guess the suggestion is that to make the patch consistent, preceding capabilities should be detected and handled. I am not sure what the best way to do it would be. Perhaps scanning through the chain until
0x09is found and reporting its offset through0x34instead of what is there would be the way to go. Then unchain anything past the0x09caps too as is currently done.
Quoted 4 lines
Or just scan through the capabilities, and chain only the ones that we want to "Whitelist" and the rest are to be blacklisted. The rest can also have its values set to some bogus value (
0xdeadbeef?) Perhaps only when built with 'debug=y'.
That sounds about right. Back when I first did the patch (in an old qemu) there were no other capabilities on the piix4 host bridge so it was simple. Not sure if other capabilities will be an issue now.
Quoted 4 lines
Anyhow, if we retain the location (aka offset) of the capabilities then if there is some silly code that expects those to be at specific locations it should still work?
Yes, that could be a definite concern. When I was dredging up memories of finding the issue, I had a notion that this might actually be the case. That is why I suggest leaving them at their offset.
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 74 lines · Ross Philipson
Quoted 67 lines · Konrad Rzeszutek Wilk
Quoted 45 lines · Ross Philipson
Quoted 34 lines · Konrad Rzeszutek Wilk
Quoted 9 lines · G.R.
I'm going to rework this patch to address Jan's concern. Here is my proposal, please review and comment before I begin:
The proposal is to read a shadow copy of the exposed host register into the config space of the emulated host bridge and relies on the
pci_default_read_config()function
to provide proper access.Quoted 6 lines
This methodology only works for constant values, which is our case here.
The exposed value is either read-only or write-locked (for BIOS).The only exception is that the PAVPC (
0x58) register is write-locked but not for BIOS.So only SeaBIOS or hvmloader should touch it?
Quoted 2 lines
This is exposed for RW and my proposal is to perform write-through in the register write-support.
What does PAVPC do? As in if the driver wrote
0xdeadbeefin there what would happen? Is there a list of the appropiate values it should accept?Quoted 5 lines
Quoted 4 lines
Also, why would removing the next capability be correct here, when you're not removing _all_ other capabilities?
Quoted 7 lines
Quoted 2 lines
I have no answer about this question. *Jean*, could you help comment since this is from your code?
If he doesn't answer - if you don't remove the capability does it still work?
Quoted 6 lines
So I actually originally found this issue with the vendor capabilities and created the original patch for it. This was quite some time ago so I had to go back and look. IIRC the vendor specific capabilities were always the first one in the chain and the unchaining code would drop all further capabilities (which we did not want to pass directly to the guest).
OK, so blacklisting.
Quoted 9 lines
We never saw a configuration where the vendor capabilities were not the first. I guess the suggestion is that to make the patch consistent, preceding capabilities should be detected and handled. I am not sure what the best way to do it would be. Perhaps scanning through the chain until
0x09is found and reporting its offset through0x34instead of what is there would be the way to go. Then unchain anything past the0x09caps too as is currently done.Quoted 4 lines
Or just scan through the capabilities, and chain only the ones that we want to "Whitelist" and the rest are to be blacklisted. The rest can also have its values set to some bogus value (
0xdeadbeef?) Perhaps only when built with 'debug=y'.
Quoted 3 lines
That sounds about right. Back when I first did the patch (in an old qemu) there were no other capabilities on the piix4 host bridge so it was simple. Not sure if other capabilities will be an issue now.
It's still the case as for the IVB host bridge.
And from what I can find from the datasheet for the Haswell, it's still the case.
Note that the datasheet explicitly documents the offset of the CAPABILITY registers.
I guess there will be code that rely on this offset that been publicly documented.
Btw. Ross, now that you appears to be the original author (sorry for mess you up with Jean),
could you also comment on my rework proposal? Jan believe the current form is not clean enough.
Currently we use a whitelist of registers to pass-through.How do you come up with the current list?
The shadow copy way appears to work for the current list. But what if we are going to need some special registers that cannot be handled well? (e.g. has side effect for reading and cannot perform read-back?)
Thanks,
Timothy
Quoted 11 lines
Quoted 6 lines
Anyhow, if we retain the location (aka offset) of the capabilities then if there is some silly code that expects those to be at specific locations it should still work?
Yes, that could be a definite concern. When I was dredging up memories of finding the issue, I had a notion that this might actually be the case. That is why I suggest leaving them at their offset.
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 89 lines · G.R.
Quoted 74 lines · Ross Philipson
Quoted 67 lines · Konrad Rzeszutek Wilk
Quoted 45 lines · Ross Philipson
Quoted 34 lines · Konrad Rzeszutek Wilk
Quoted 9 lines · G.R.
I'm going to rework this patch to address Jan's concern. Here is my proposal, please review and comment before I begin:
The proposal is to read a shadow copy of the exposed host register into the config space of the emulated host bridge and relies on the
pci_default_read_config()function
to provide proper access.Quoted 6 lines
This methodology only works for constant values, which is our case here.
The exposed value is either read-only or write-locked (for BIOS).The only exception is that the PAVPC (
0x58) register is write-locked but not for BIOS.So only SeaBIOS or hvmloader should touch it?
Quoted 2 lines
This is exposed for RW and my proposal is to perform write-through in the register write-support.
What does PAVPC do? As in if the driver wrote
0xdeadbeefin there what would happen? Is there a list of the appropiate values it should accept?Quoted 5 lines
Quoted 4 lines
Also, why would removing the next capability be correct here, when you're not removing _all_ other capabilities?
Quoted 7 lines
Quoted 2 lines
I have no answer about this question. *Jean*, could you help comment since this is from your code?
If he doesn't answer - if you don't remove the capability does it still work?
Quoted 6 lines
So I actually originally found this issue with the vendor capabilities and created the original patch for it. This was quite some time ago so I had to go back and look. IIRC the vendor specific capabilities were always the first one in the chain and the unchaining code would drop all further capabilities (which we did not want to pass directly to the guest).
OK, so blacklisting.
Quoted 9 lines
We never saw a configuration where the vendor capabilities were not the first. I guess the suggestion is that to make the patch consistent, preceding capabilities should be detected and handled. I am not sure what the best way to do it would be. Perhaps scanning through the chain until
0x09is found and reporting its offset through0x34instead of what is there would be the way to go. Then unchain anything past the0x09caps too as is currently done.Quoted 4 lines
Or just scan through the capabilities, and chain only the ones that we want to "Whitelist" and the rest are to be blacklisted. The rest can also have its values set to some bogus value (
0xdeadbeef?) Perhaps only when built with 'debug=y'.Quoted 3 lines
That sounds about right. Back when I first did the patch (in an old qemu) there were no other capabilities on the piix4 host bridge so it was simple. Not sure if other capabilities will be an issue now.
It's still the case as for the IVB host bridge.
And from what I can find from the datasheet for the Haswell, it's still the case.Note that the datasheet explicitly documents the offset of the CAPABILITY registers.
I guess there will be code that rely on this offset that been publicly documented.
Quoted 8 lines
Btw. Ross, now that you appears to be the original author (sorry for mess you up with Jean),
could you also comment on my rework proposal? Jan believe the current form is not clean enough.Currently we use a whitelist of registers to pass-through.How do you come up with the current list?
The shadow copy way appears to work for the current list.
OK.
Quoted 3 lines
But what if we are going to need some special registers that cannot be handled well? (e.g. has side effect for reading and cannot perform read-back?)
Hopefully the i915 driver in Linux will help in figuring out which ones of those are needed?
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 81 lines · Konrad Rzeszutek Wilk
Quoted 80 lines · G.R.
Quoted 77 lines · Ross Philipson
Quoted 67 lines · Konrad Rzeszutek Wilk
Quoted 45 lines · Ross Philipson
Quoted 34 lines · Konrad Rzeszutek Wilk
Quoted 9 lines · G.R.
I'm going to rework this patch to address Jan's concern. Here is my proposal, please review and comment before I begin:
The proposal is to read a shadow copy of the exposed host register into the config space of the emulated host bridge and relies on the
pci_default_read_config()function
to provide proper access.Quoted 6 lines
This methodology only works for constant values, which is our case here.
The exposed value is either read-only or write-locked (for BIOS).The only exception is that the PAVPC (
0x58) register is write-locked but not for BIOS.So only SeaBIOS or hvmloader should touch it?
Quoted 2 lines
This is exposed for RW and my proposal is to perform write-through in the register write-support.
What does PAVPC do? As in if the driver wrote
0xdeadbeefin there what would happen? Is there a list of the appropiate values it should accept?Quoted 5 lines
Quoted 4 lines
Also, why would removing the next capability be correct here, when you're not removing _all_ other capabilities?
Quoted 7 lines
Quoted 2 lines
I have no answer about this question. *Jean*, could you help comment since this is from your code?
If he doesn't answer - if you don't remove the capability does it still work?
Quoted 6 lines
So I actually originally found this issue with the vendor capabilities and created the original patch for it. This was quite some time ago so I had to go back and look. IIRC the vendor specific capabilities were always the first one in the chain and the unchaining code would drop all further capabilities (which we did not want to pass directly to the guest).
OK, so blacklisting.
Quoted 9 lines
We never saw a configuration where the vendor capabilities were not the first. I guess the suggestion is that to make the patch consistent, preceding capabilities should be detected and handled. I am not sure what the best way to do it would be. Perhaps scanning through the chain until
0x09is found and reporting its offset through0x34instead of what is there would be the way to go. Then unchain anything past the0x09caps too as is currently done.Quoted 4 lines
Or just scan through the capabilities, and chain only the ones that we want to "Whitelist" and the rest are to be blacklisted. The rest can also have its values set to some bogus value (
0xdeadbeef?) Perhaps only when built with 'debug=y'.That sounds about right. Back when I first did the patch (in an old qemu) there were no other capabilities on the piix4 host bridge so it was simple. Not sure if other capabilities will be an issue now.
Quoted 8 lines
Quoted 8 lines
It's still the case as for the IVB host bridge.
And from what I can find from the datasheet for the Haswell, it's still the case.Note that the datasheet explicitly documents the offset of the CAPABILITY registers.
I guess there will be code that rely on this offset that been publicly documented.
IIRC I think that may be the case; probably should proceed with that assumption.
Quoted 3 lines
Quoted 3 lines
Btw. Ross, now that you appears to be the original author (sorry for mess you up with Jean),
No worries - Jean and I used to work together. He took the patch, cleaned it up and upstreamed it.
Quoted 2 lines
Quoted 2 lines
could you also comment on my rework proposal? Jan believe the current form is not clean enough.
Using the shadow registers? That sounds fine to me.
Quoted 4 lines
Quoted 4 lines
Currently we use a whitelist of registers to pass-through.How do you come up with the current list?
The shadow copy way appears to work for the current list.
Originally the only registers being dealt with were the vendor capabilities. The other registers came along later and I am not sure who identified them or why.
For example, I looked up the PAVPC register and it is pretty much undocumented in the spec I have. It says it is disabled by TXT which I assume means if you don't have TXT (or it is disabled), this register would not be disabled for write. I am not sure the best way to handle all that. Maybe set the PAVPLCK bit at the appropriate time and just disable it.
Quoted 8 lines
OK.
Quoted 3 lines
But what if we are going to need some special registers that cannot be handled well? (e.g. has side effect for reading and cannot perform read-back?)
Hopefully the i915 driver in Linux will help in figuring out which ones of those are needed?
That sounds like a good suggestion since a lot of those registers are minimally documented.
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
In reply to Konrad Rzeszutek WilkQuoted 99 lines · Konrad Rzeszutek Wilk
Quoted 89 lines · G.R.
Quoted 74 lines · Ross Philipson
Quoted 67 lines · Konrad Rzeszutek Wilk
Quoted 45 lines · Ross Philipson
Quoted 34 lines · Konrad Rzeszutek Wilk
Quoted 9 lines · G.R.
I'm going to rework this patch to address Jan's concern. Here is my proposal, please review and comment before I begin:
The proposal is to read a shadow copy of the exposed host register into the config space of the emulated host bridge and relies on the
pci_default_read_config()function
to provide proper access.Quoted 6 lines
This methodology only works for constant values, which is our case here.
The exposed value is either read-only or write-locked (for BIOS).The only exception is that the PAVPC (
0x58) register is write-locked but not for BIOS.So only SeaBIOS or hvmloader should touch it?
Quoted 2 lines
This is exposed for RW and my proposal is to perform write-through in the register write-support.
What does PAVPC do? As in if the driver wrote
0xdeadbeefin there what would happen? Is there a list of the appropiate values it should accept?Quoted 5 lines
Quoted 4 lines
Also, why would removing the next capability be correct here, when you're not removing _all_ other capabilities?
Quoted 7 lines
Quoted 2 lines
I have no answer about this question. *Jean*, could you help comment since this is from your code?
If he doesn't answer - if you don't remove the capability does it still work?
Quoted 6 lines
So I actually originally found this issue with the vendor capabilities and created the original patch for it. This was quite some time ago so I had to go back and look. IIRC the vendor specific capabilities were always the first one in the chain and the unchaining code would drop all further capabilities (which we did not want to pass directly to the guest).
OK, so blacklisting.
Quoted 9 lines
We never saw a configuration where the vendor capabilities were not the first. I guess the suggestion is that to make the patch consistent, preceding capabilities should be detected and handled. I am not sure what the best way to do it would be. Perhaps scanning through the chain until
0x09is found and reporting its offset through0x34instead of what is there would be the way to go. Then unchain anything past the0x09caps too as is currently done.Quoted 4 lines
Or just scan through the capabilities, and chain only the ones that we want to "Whitelist" and the rest are to be blacklisted. The rest can also have its values set to some bogus value (
0xdeadbeef?) Perhaps only when built with 'debug=y'.Quoted 3 lines
That sounds about right. Back when I first did the patch (in an old qemu) there were no other capabilities on the piix4 host bridge so it was simple. Not sure if other capabilities will be an issue now.
It's still the case as for the IVB host bridge.
And from what I can find from the datasheet for the Haswell, it's still the case.Note that the datasheet explicitly documents the offset of the CAPABILITY registers.
I guess there will be code that rely on this offset that been publicly documented.Quoted 8 lines
Btw. Ross, now that you appears to be the original author (sorry for mess you up with Jean),
could you also comment on my rework proposal? Jan believe the current form is not clean enough.Currently we use a whitelist of registers to pass-through.How do you come up with the current list?
The shadow copy way appears to work for the current list.
Quoted 7 lines
OK.
Quoted 3 lines
But what if we are going to need some special registers that cannot be handled well? (e.g. has side effect for reading and cannot perform read-back?)
Hopefully the i915 driver in Linux will help in figuring out which ones of those are needed?
I remember the vendor cap fix only helps windows guest. Linux guest just run happily without this.
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 28 lines
Quoted 28 lines
Quoted 27 lines
Quoted 9 lines
Quoted 4 lines
Or just scan through the capabilities, and chain only the ones that we want to "Whitelist" and the rest are to be blacklisted. The rest can also have its values set to some bogus value (
0xdeadbeef?) Perhaps only when built with 'debug=y'.That sounds about right. Back when I first did the patch (in an old qemu) there were no other capabilities on the piix4 host bridge so it was simple. Not sure if other capabilities will be an issue now.
It's still the case as for the IVB host bridge.
And from what I can find from the datasheet for the Haswell, it's still the case.Note that the datasheet explicitly documents the offset of the CAPABILITY registers.
I guess there will be code that rely on this offset that been publicly documented.Btw. Ross, now that you appears to be the original author (sorry for mess you up with Jean),
could you also comment on my rework proposal? Jan believe the current form is not clean enough.Currently we use a whitelist of registers to pass-through.How do you come up with the current list?
The shadow copy way appears to work for the current list.
Quoted 8 lines
Quoted 7 lines
OK.
Quoted 3 lines
But what if we are going to need some special registers that cannot be handled well? (e.g. has side effect for reading and cannot perform read-back?)
Hopefully the i915 driver in Linux will help in figuring out which ones of those are needed?
I remember the vendor cap fix only helps windows guest.
How was that diagnosed? Perhaps that information can be part of the source code to help in the future with diagnosiing which caps are needed and which ones can be blacklisted?
Quoted 7 lines
Linux guest just run happily without this.
_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 33 lines · Konrad Rzeszutek Wilk
Quoted 33 lines
Quoted 33 lines
Quoted 19 lines
Quoted 9 lines
Quoted 4 lines
Or just scan through the capabilities, and chain only the ones that we want to "Whitelist" and the rest are to be blacklisted. The rest can also have its values set to some bogus value (
0xdeadbeef?) Perhaps only when built with 'debug=y'.That sounds about right. Back when I first did the patch (in an old qemu) there were no other capabilities on the piix4 host bridge so it was simple. Not sure if other capabilities will be an issue now.
It's still the case as for the IVB host bridge.
And from what I can find from the datasheet for the Haswell, it's still the case.Note that the datasheet explicitly documents the offset of the CAPABILITY registers.
I guess there will be code that rely on this offset that been publicly documented.Quoted 8 lines
Btw. Ross, now that you appears to be the original author (sorry for mess you up with Jean),
could you also comment on my rework proposal? Jan believe the current form is not clean enough.Currently we use a whitelist of registers to pass-through.How do you come up with the current list?
The shadow copy way appears to work for the current list.OK.
Quoted 3 lines
But what if we are going to need some special registers that cannot be handled well? (e.g. has side effect for reading and cannot perform read-back?)
Quoted 8 lines
Quoted 3 lines
Quoted 2 lines
Hopefully the i915 driver in Linux will help in figuring out which ones of those are needed?
I remember the vendor cap fix only helps windows guest.
How was that diagnosed? Perhaps that information can be part of the source code to help in the future with diagnosiing which caps are needed and which ones can be blacklisted?
I guess that's a question mostly for Ross/Jean as they're the original authors of the patch?
-- Pasi
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 36 lines · Pasi Kärkkäinen
Quoted 35 lines · Konrad Rzeszutek Wilk
Quoted 35 lines
Quoted 35 lines
Quoted 19 lines
Quoted 9 lines
Quoted 4 lines
Or just scan through the capabilities, and chain only the ones that we want to "Whitelist" and the rest are to be blacklisted. The rest can also have its values set to some bogus value (
0xdeadbeef?) Perhaps only when built with 'debug=y'.That sounds about right. Back when I first did the patch (in an old qemu) there were no other capabilities on the piix4 host bridge so it was simple. Not sure if other capabilities will be an issue now.
It's still the case as for the IVB host bridge.
And from what I can find from the datasheet for the Haswell, it's still the case.Note that the datasheet explicitly documents the offset of the CAPABILITY registers.
I guess there will be code that rely on this offset that been publicly documented.Quoted 8 lines
Btw. Ross, now that you appears to be the original author (sorry for mess you up with Jean),
could you also comment on my rework proposal? Jan believe the current form is not clean enough.Currently we use a whitelist of registers to pass-through.How do you come up with the current list?
The shadow copy way appears to work for the current list.OK.
Quoted 3 lines
But what if we are going to need some special registers that cannot be handled well? (e.g. has side effect for reading and cannot perform read-back?)
Hopefully the i915 driver in Linux will help in figuring out which ones of those are needed?
Quoted 8 lines
Quoted 6 lines
Quoted 1 line
I remember the vendor cap fix only helps windows guest.
How was that diagnosed? Perhaps that information can be part of the source code to help in the future with diagnosiing which caps are needed and which ones can be blacklisted?
I guess that's a question mostly for Ross/Jean as they're the original authors of the patch?
We discovered the issue with Windows guests running the vendor drivers for the passed in IGD graphics device. Under certain circumstances (resuming from S3/S4 IIRC), the guest would BSOD. I finally tracked it down to a bad state in the resuming driver because it was not coded to handle the vendor capabilities not being present on the host bridge. BTW, those capabilities are flags indicating what features the IGD card has - their exact meaning is of course proprietary.
I cannot say it was only a problem on Windows but rather that that is the only place we ever saw it.
I never saw any other capabilities on the hosts bridges at that time, just vendor ones so the patch just handled that. If there were other capabilities, I would think it would have to be determined on a case by case basis whether they were included. Inclusion of each new type would have different ramifications it seems.
Quoted 3 lines
-- Pasi
_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 55 lines · Ross Philipson
Quoted 36 lines · Pasi Kärkkäinen
Quoted 35 lines · Konrad Rzeszutek Wilk
Quoted 35 lines
Quoted 35 lines
Quoted 19 lines
Quoted 9 lines
Quoted 4 lines
Or just scan through the capabilities, and chain only the ones that we want to "Whitelist" and the rest are to be blacklisted. The rest can also have its values set to some bogus value (
0xdeadbeef?) Perhaps only when built with 'debug=y'.That sounds about right. Back when I first did the patch (in an old qemu) there were no other capabilities on the piix4 host bridge so it was simple. Not sure if other capabilities will be an issue now.
It's still the case as for the IVB host bridge.
And from what I can find from the datasheet for the Haswell, it's still the case.Note that the datasheet explicitly documents the offset of the CAPABILITY registers.
I guess there will be code that rely on this offset that been publicly documented.Quoted 8 lines
Btw. Ross, now that you appears to be the original author (sorry for mess you up with Jean),
could you also comment on my rework proposal? Jan believe the current form is not clean enough.Currently we use a whitelist of registers to pass-through.How do you come up with the current list?
The shadow copy way appears to work for the current list.OK.
Quoted 3 lines
But what if we are going to need some special registers that cannot be handled well? (e.g. has side effect for reading and cannot perform read-back?)
Hopefully the i915 driver in Linux will help in figuring out which ones of those are needed?
Quoted 8 lines
Quoted 6 lines
Quoted 1 line
I remember the vendor cap fix only helps windows guest.
How was that diagnosed? Perhaps that information can be part of the source code to help in the future with diagnosiing which caps are needed and which ones can be blacklisted?
I guess that's a question mostly for Ross/Jean as they're the original authors of the patch?
We discovered the issue with Windows guests running the vendor drivers for the passed in IGD graphics device. Under certain circumstances (resuming from S3/S4 IIRC), the guest would BSOD. I finally tracked it down to a bad state in the resuming driver because it was not coded to handle the vendor capabilities not being present on the host bridge. BTW, those capabilities are flags indicating what features the IGD card has - their exact meaning is of course proprietary.
Quoted 9 lines
I cannot say it was only a problem on Windows but rather that that is the only place we ever saw it.
I never saw any other capabilities on the hosts bridges at that time, just vendor ones so the patch just handled that. If there were other capabilities, I would think it would have to be determined on a case by case basis whether they were included. Inclusion of each new type would have different ramifications it seems.
Thanks for the explanation. I guess parts of that should go to the patch description aswell..
-- Pasi
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 19 lines · Pasi Kärkkäinen
Quoted 19 lines
Quoted 6 lines
Quoted 4 lines
How was that diagnosed? Perhaps that information can be part of the source code to help in the future with diagnosiing which caps are needed and which ones can be blacklisted?
I guess that's a question mostly for Ross/Jean as they're the original authors of the patch?
We discovered the issue with Windows guests running the vendor drivers for the passed in IGD graphics device. Under certain circumstances (resuming from S3/S4 IIRC), the guest would BSOD. I finally tracked it down to a bad state in the resuming driver because it was not coded to handle the vendor capabilities not being present on the host bridge. BTW, those capabilities are flags indicating what features the IGD card has - their exact meaning is of course proprietary.
I cannot say it was only a problem on Windows but rather that that is the only place we ever saw it.
Quoted 9 lines
Quoted 6 lines
I never saw any other capabilities on the hosts bridges at that time, just vendor ones so the patch just handled that. If there were other capabilities, I would think it would have to be determined on a case by case basis whether they were included. Inclusion of each new type would have different ramifications it seems.
Thanks for the explanation. I guess parts of that should go to the patch description aswell..
Now patch 2/3 has been applied to qemu-traditional, so only this patch 3/3 is missing from qemu-traditional from this series.
Any other changes outstanding? or only to add some of Ross's comments to the patch description (and/or sources) ?
Thanks,
-- Pasi
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Quoted 27 lines · Pasi Kärkkäinen
Quoted 26 lines · Pasi Kärkkäinen
Quoted 25 lines
Quoted 6 lines
Quoted 4 lines
How was that diagnosed? Perhaps that information can be part of the source code to help in the future with diagnosiing which caps are needed and which ones can be blacklisted?
I guess that's a question mostly for Ross/Jean as they're the original authors of the patch?
We discovered the issue with Windows guests running the vendor drivers for the passed in IGD graphics device. Under certain circumstances (resuming from S3/S4 IIRC), the guest would BSOD. I finally tracked it down to a bad state in the resuming driver because it was not coded to handle the vendor capabilities not being present on the host bridge. BTW, those capabilities are flags indicating what features the IGD card has - their exact meaning is of course proprietary.
I cannot say it was only a problem on Windows but rather that that is the only place we ever saw it.
I never saw any other capabilities on the hosts bridges at that time, just vendor ones so the patch just handled that. If there were other capabilities, I would think it would have to be determined on a case by case basis whether they were included. Inclusion of each new type would have different ramifications it seems.
Quoted 7 lines
Quoted 2 lines
Thanks for the explanation. I guess parts of that should go to the patch description aswell..
Now patch 2/3 has been applied to qemu-traditional, so only this patch 3/3 is missing from qemu-traditional from this series.
Any other changes outstanding? or only to add some of Ross's comments to the patch description (and/or sources) ?
I'll have to rework this patch -- Jan believe the code is not very clean with regard to offset / size handling.
I proposed an alternative to shadow the registers into the PCI config space of the emulated host controller.
There seems no objection on this proposal. I'll do it when I got some spare time.
Quoted 4 lines
Thanks,
-- Pasi
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
Hello,
Quoted 5 lines · G.R.
Quoted 5 lines
Now patch 2/3 has been applied to qemu-traditional, so only this patch 3/3 is missing from qemu-traditional from this series.
Any other changes outstanding? or only to add some of Ross's comments to the patch description (and/or sources) ?
Quoted 8 lines
I'll have to rework this patch -- Jan believe the code is not very clean with regard to offset / size handling.
I proposed an alternative to shadow the registers into the PCI config space of the emulated host controller.
There seems no objection on this proposal. I'll do it when I got some spare time.
Do you think you'll find time to rework patch 3/3 before Xen 4.4 feature freeze? It'd be nice to get the last patch from this series finally applied!
Thanks,
-- Pasi
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
In reply to G.R.Hello Tiejun,
Since you've been working on upstreaming the Xen IGD passthru patches to qemu-upstream, you might be able to comment on this patch aswell.
This patch is for qemu-traditional, but this one hasn't been acked / applied yet. It has been floating around on xen-devel for a couple of years now. It would be nice to get this finally merged to qemu-traditional..
Can you comment about best approach to implement this patch? There was some comments/concerns about the earlier versions.
"[PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.":
http://lists.xenproject.org/archives/html/xen-devel/2013-02/msg00538.html
And some discussion about the patch/justification one year ago:
http://lists.xen.org/archives/html/xen-devel/2013-07/msg01376.html
Also G.R. wrote earlier on this thread:
Quoted 3 lines
I proposed an alternative to shadow the registers into the PCI config space of the emulated host controller.
There seems no objection on this proposal. I'll do it when I got some spare time.
Thanks,
-- Pasi
Quoted 47 lines · G.R.
Quoted 27 lines · Pasi Kärkkäinen
Quoted 26 lines · Pasi Kärkkäinen
Quoted 25 lines
Quoted 6 lines
Quoted 4 lines
How was that diagnosed? Perhaps that information can be part of the source code to help in the future with diagnosiing which caps are needed and which ones can be blacklisted?
I guess that's a question mostly for Ross/Jean as they're the original authors of the patch?
We discovered the issue with Windows guests running the vendor drivers for the passed in IGD graphics device. Under certain circumstances (resuming from S3/S4 IIRC), the guest would BSOD. I finally tracked it down to a bad state in the resuming driver because it was not coded to handle the vendor capabilities not being present on the host bridge. BTW, those capabilities are flags indicating what features the IGD card has - their exact meaning is of course proprietary.
I cannot say it was only a problem on Windows but rather that that is the only place we ever saw it.
I never saw any other capabilities on the hosts bridges at that time, just vendor ones so the patch just handled that. If there were other capabilities, I would think it would have to be determined on a case by case basis whether they were included. Inclusion of each new type would have different ramifications it seems.
Quoted 7 lines
Quoted 2 lines
Thanks for the explanation. I guess parts of that should go to the patch description aswell..
Now patch 2/3 has been applied to qemu-traditional, so only this patch 3/3 is missing from qemu-traditional from this series.
Any other changes outstanding? or only to add some of Ross's comments to the patch description (and/or sources) ?
I'll have to rework this patch -- Jan believe the code is not very clean with regard to offset / size handling.
I proposed an alternative to shadow the registers into the PCI config space of the emulated host controller.
There seems no objection on this proposal. I'll do it when I got some spare time.Quoted 4 lines
Thanks,
-- Pasi
Re: [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific pci cap on host bridge.
In reply to Konrad Rzeszutek WilkQuoted 9 lines · Konrad Rzeszutek Wilk
Quoted 8 lines · G.R.
I'm going to rework this patch to address Jan's concern. Here is my proposal, please review and comment before I begin:
The proposal is to read a shadow copy of the exposed host register into the config space of the emulated host bridge and relies on the
pci_default_read_config()function
to provide proper access.
Quoted 8 lines
Quoted 5 lines
This methodology only works for constant values, which is our case here. The exposed value is either read-only or write-locked (for BIOS).
The only exception is that the PAVPC (
0x58) register is write-locked but not for BIOS.So only SeaBIOS or hvmloader should touch it?
No, here I mean the host BIOS.
Those write-locked registers should have been locked by host BIOS and be read-only after boot.
Most of these are for graphics memory. I'm not sure how the value specified by host BIOS works in the VM, but it just works.
You can find a list of these registers in this document (section 2.5, page 47):
http://www.intel.com/content/dam/www/public/us/en/documents/datasheets/3rd-gen-core-desktop-vol-2-datasheet.pdf
Just for your reference, here are the list of registers that exposed RO (with the exception of 0x58 as RW):
198 case 0x00: /* vendor id */ 199 case 0x02: /* device id */ 201 case 0x06: /* status, needed for the cap list bit*/ 202 case 0x08: /* revision id */ 203 case 0x2c: /* sybsystem vendor id */ 204 case 0x2e: /* sybsystem id */ 205 case 0x50: /* SNB: processor graphics control register */ 206 case 0x52: /* processor graphics control register */ <= actually RSVD from datasheet 207 case 0xa0: /* top of memory */ 208 case 0xb0: /* ILK: BSM: should read from dev 2 offset 0x5c */ 209 case 0x58: /* SNB: PAVPC Offset */ 210 case 0xa4: /* SNB: graphics base of stolen memory */ 211 case 0xa8: /* SNB: base of GTT stolen memory */
Quoted 7 lines
Quoted 2 lines
This is exposed for RW and my proposal is to perform write-through in the register write-support.
What does PAVPC do? As in if the driver wrote
0xdeadbeefin there what would happen? Is there a list of the appropiate values it should accept?
I have no idea about this.
I can't get meaningful data from the datasheet.
And the value returned from lspci can't decode well according to the datasheet. The datasheet above shows only one write-lock bit at bit 2, while all others are RO and reset to zero.
But here is the lspci value from my system (The four bytes starting from 0x58):
50: 41 02 00 00 11 00 00 00 07 00 90 df 01 00 00 cf
Anyway, I don't think we should dig into the detailed register spec. The good news is that none of the registers in device 0:00.0 have side effect for read.
We should be able to perform write to host and read-back to shadow for RW support.
Quoted 10 lines
Quoted 6 lines
Quoted 4 lines
Also, why would removing the next capability be correct here, when you're not removing _all_ other capabilities?
I have no answer about this question. *Jean*, could you help comment since this is from your code?
If he doesn't answer - if you don't remove the capability does it still work?
Should work at least for IVB -- since this is the only cap. Not sure if there will be other concerns.
Quoted 14 lines
Quoted 9 lines
Thanks,
Timothy_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel
Re: Patch series for IGD passthrough
In reply to Rui GuoQuoted 9 lines · Rui Guo
This series contains all the fixes required to produce a working IGD passthrough box. All the changes are previously seen in the dev list but not yet accepted. Some of them are out-dated and need some reshape.
Detailed description can be found later in each patch.
. [PATCH 1/3] qemu-xen-trad/pt_msi_disable: do not clear all MSI flags . [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD . [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific
All patches look good
Re: Patch series for IGD passthrough
In reply to Rui GuoQuoted 10 lines · Rui Guo
This series contains all the fixes required to produce a working IGD passthrough box. All the changes are previously seen in the dev list but not yet accepted. Some of them are out-dated and need some reshape.
Detailed description can be found later in each patch.
. [PATCH 1/3] qemu-xen-trad/pt_msi_disable: do not clear all MSI flags . [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD . [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific
Looking at qemu-xen-unstable I think patches 2 and 3 are not yet applied. (patch 2 was applied earlier but it got reverted).
Is it likely that we'll get these merged in for Xen 4.3 ?
Thanks,
-- Pasi
Re: Patch series for IGD passthrough
Quoted 12 lines · Pasi Kärkkäinen
Quoted 10 lines · Rui Guo
This series contains all the fixes required to produce a working IGD passthrough box. All the changes are previously seen in the dev list but not yet accepted. Some of them are out-dated and need some reshape.
Detailed description can be found later in each patch.
. [PATCH 1/3] qemu-xen-trad/pt_msi_disable: do not clear all MSI flags . [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD . [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific
Quoted 5 lines
Looking at qemu-xen-unstable I think patches 2 and 3 are not yet applied. (patch 2 was applied earlier but it got reverted).
Is it likely that we'll get these merged in for Xen 4.3 ?
Ping?
-- Pasi
Re: Patch series for IGD passthrough
Quoted 13 lines · Pasi Kärkkäinen
Quoted 12 lines · Pasi Kärkkäinen
Quoted 10 lines · Rui Guo
This series contains all the fixes required to produce a working IGD passthrough box. All the changes are previously seen in the dev list but not yet accepted. Some of them are out-dated and need some reshape.
Detailed description can be found later in each patch.
. [PATCH 1/3] qemu-xen-trad/pt_msi_disable: do not clear all MSI flags . [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD . [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific
Quoted 7 lines
Quoted 5 lines
Looking at qemu-xen-unstable I think patches 2 and 3 are not yet applied. (patch 2 was applied earlier but it got reverted).
Is it likely that we'll get these merged in for Xen 4.3 ?
Ping?
Whose decision is this?
I thought we were actually not going to accept new functionality into qemu-trad?
-George
Re: Patch series for IGD passthrough
Quoted 17 lines · George Dunlap
Quoted 16 lines · Pasi Kärkkäinen
Quoted 15 lines · Pasi Kärkkäinen
Quoted 10 lines · Rui Guo
This series contains all the fixes required to produce a working IGD passthrough box. All the changes are previously seen in the dev list but not yet accepted. Some of them are out-dated and need some reshape.
Detailed description can be found later in each patch.
. [PATCH 1/3] qemu-xen-trad/pt_msi_disable: do not clear all MSI flags . [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD . [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific
Looking at qemu-xen-unstable I think patches 2 and 3 are not yet applied. (patch 2 was applied earlier but it got reverted).
Quoted 7 lines
Quoted 4 lines
Quoted 2 lines
Is it likely that we'll get these merged in for Xen 4.3 ?
Ping?
Whose decision is this?
I think at least one of the patches needs fixing and re-send..
Quoted 2 lines
I thought we were actually not going to accept new functionality into qemu-trad?
I don't think this is a new feature really.. it's just fixing the existing Intel VGA passthru functionality.
-- Pasi
Re: Patch series for IGD passthrough
In reply to George DunlapQuoted 17 lines · George Dunlap
Quoted 16 lines · Pasi Kärkkäinen
Quoted 15 lines · Pasi Kärkkäinen
Quoted 10 lines · Rui Guo
This series contains all the fixes required to produce a working IGD passthrough box. All the changes are previously seen in the dev list but not yet accepted. Some of them are out-dated and need some reshape.
Detailed description can be found later in each patch.
. [PATCH 1/3] qemu-xen-trad/pt_msi_disable: do not clear all MSI flags . [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD . [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific
Looking at qemu-xen-unstable I think patches 2 and 3 are not yet applied. (patch 2 was applied earlier but it got reverted).
Quoted 6 lines
Quoted 4 lines
Quoted 2 lines
Is it likely that we'll get these merged in for Xen 4.3 ?
Ping?
Whose decision is this?
Ian Jackson
Quoted 1 line
I thought we were actually not going to accept new functionality into qemu-trad?
They are fixes
_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel
Re: Patch series for IGD passthrough
In reply to Pasi KärkkäinenQuoted 13 lines · Pasi Kärkkäinen
Quoted 12 lines · Pasi Kärkkäinen
Quoted 10 lines · Rui Guo
This series contains all the fixes required to produce a working IGD passthrough box. All the changes are previously seen in the dev list but not yet accepted. Some of them are out-dated and need some reshape.
Detailed description can be found later in each patch.
. [PATCH 1/3] qemu-xen-trad/pt_msi_disable: do not clear all MSI flags . [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD . [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific
Quoted 7 lines
Quoted 5 lines
Looking at qemu-xen-unstable I think patches 2 and 3 are not yet applied. (patch 2 was applied earlier but it got reverted).
Is it likely that we'll get these merged in for Xen 4.3 ?
Ping?
That's a call for Ian Jackson, CC'ing him.
_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel
Re: Patch series for IGD passthrough
IanJ: ping again..
-- Pasi
Quoted 23 lines · Stefano Stabellini
Quoted 13 lines · Pasi Kärkkäinen
Quoted 12 lines · Pasi Kärkkäinen
Quoted 10 lines · Rui Guo
This series contains all the fixes required to produce a working IGD passthrough box. All the changes are previously seen in the dev list but not yet accepted. Some of them are out-dated and need some reshape.
Detailed description can be found later in each patch.
. [PATCH 1/3] qemu-xen-trad/pt_msi_disable: do not clear all MSI flags . [PATCH 2/3] qemu-xen-trad: Correctly expose PCH ISA bridge for IGD . [PATCH 3/3] qemu-xen-trad: IGD passthrough: Expose vendor specific
Quoted 7 lines
Quoted 5 lines
Looking at qemu-xen-unstable I think patches 2 and 3 are not yet applied. (patch 2 was applied earlier but it got reverted).
Is it likely that we'll get these merged in for Xen 4.3 ?
Ping?
That's a call for Ian Jackson, CC'ing him.