-
Notifications
You must be signed in to change notification settings - Fork 879
Conversation
core/ept2.c
Outdated
if (combined_perm != HAX_EPT_PERM_NONE) { | ||
if ((qual.raw & HAX_EPT_ACC_W) && !(combined_perm & HAX_EPT_PERM_W) && | ||
(slot->flags == HAX_MEMSLOT_READONLY)) { | ||
// write to rom device |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Write to ROM device
@@ -45,9 +45,14 @@ | |||
#define HAX_EPT_TABLE_SHIFT 9 | |||
#define HAX_EPT_TABLE_SIZE (1 << HAX_EPT_TABLE_SHIFT) | |||
|
|||
#define HAX_EPT_ACC_R 0x1 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
How about 'HAX_EPT_ACCESS_R'? Why define read access as 0x1 rather than 0x4?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
As ROM device is read-only, the write operation on ROM device will
generate EPT violation and then it will be handled as MMIO write,
which will cause an access error and result in VM launch failure.
9aba437
to
33ceea0
Compare
ROM device is read-only. Its read operation will pass through. Its Write operation will be handled as MMIO. Don't dinguish ROM and ROM device exactly as no side impact. Signed-off-by: Hang Yuan <hang.yuan@intel.com>
Well Intel, |
Thanks for your comments. Could you provide the detailed command line to specify the ROM for OVMF so that we can clarify the requirements further? As the main development is targeted to Android Emulator, maybe we could not cover enough test cases for QEMU. Sorry for bring you inconvenience. |
I will email details this morning. I noticed a new release yesterday and
updated to 7.5.6 with same results.
I will send the qemu command line with details. I have qemu working with
KVM under Linux. I am just finishing a book on devrloping for UEFI and
neete to cspture screens. My screen recording and capture work under
Windows. So I installed QEMU for Windows. Unfortunately, booting Linux on
Qemu under Windows performs pretty slow, so I installed haxm.
This has been under Windows 10. I will also test under Windows 7, and
Server 2012 to see if I get better results.
Thank you,
Oliver Bailey
…On Tue, Apr 7, 2020, 4:37 AM Wenchao Wang ***@***.***> wrote:
Thanks for your comments. Could you provide the detailed command line to
specify the ROM for OVMF so that we can clarify the requirements further?
As the main development is targeted to Android Emulator, maybe we could not
cover enough test cases for QEMU. Sorry for bring you inconvenience.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#213 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AD6CBICTYFLPDUZDD5N4KSDRLLX5ZANCNFSM4HY6SU4A>
.
|
Hi Oliver, "Intel is a member of the UEFI team" doesn't mandate all Intel's SW should support UEFI in default. As mentioned in @wcwang's comment, the initiative of HAXM is to support Android Emulator. So OVMF was not supported at the beginning. HAXM doesn't claim and also is impossible to support all Intel hardware features or all specifications Intel joined. Hope you can understand. Thanks, |
Let me say that I have Dell laptop with an Intel i7 processor. Given that
fact, I am running a Windows 10 64 bit version. It vould be argued that the
gnu C and C++ compilers fon't use an Intel assembler to generate Windows 64
bit binaries, but that would be a moot arguement.
I never heard of HAXM until this issue came up. With several open souce
virtualization systems that work and are mature, I had to ask why HAXM
existed in the first place.
And I'm a senior in this industry and life, and no, quite frankly I don't
understand. QEMU id a bery mature working emulator and why a simple region
of memory for a couple of ROMs cannot be set aside and work, is a big
problem.
And it sppears bssed on what I've read in this fourm, its been a problem
for some time. But a bigger problem is the atitude taken by the developers
in this fourm.
I have no idea who you are, but generally when someone does rrsch out from
Intel, they identify themselves as an Intel employee.
If you had a Chevy, and purchased a Chevy part, and it didn't work; would
you accept the same excuse you tried to hand me?
Its an Intel processor and if Intel wants to offer a hypervisor or VM
manager, it should maintaon oversight of its FOSS projects to see that it
does work.
From what I've seen in this fourm, there are several people who get
defensive and by the stats I've seen, haven't contributed anything in the
last 12 months, at least.
I'm certain Intel would have an interest, since its seems normal that using
QEMU to debug UEFI mskes perfect sense.
I'm certain I can get Intel interested in grtting this fixed, given its
importance in the overall ability to debug UEFI.
Regards,
Oliver Bailey
…On Tue, Apr 7, 2020, 11:48 AM hyuan3 ***@***.***> wrote:
Well Intel,
You developer here isn't getting the job done. You cannot specify any ROM
for OVML and boot a VM with QEMU. This developer states he signed off on
this June 2019, and everyone on the project who hasn't contributed in the
past year, gives lip service. If this problem isn't fixed, its time to
contact the escalation team, which I will do this week. There are several
people claiming to be working on this project, but only one has
contributed, and two just talk. Please put some fire under these developers
to get this fixed. Intel is a member of the UEFI team!!!
Hi Oliver, "Intel is a member of the UEFI team" doesn't mandate all
Intel's SW should support UEFI in default. As mentioned in @wcwang
<https://github.com/wcwang>'s comment, the initiative of HAXM is to
support Android Emulator. So OVMF was not supported at the beginning. HAXM
doesn't claim and also is impossible to support all Intel hardware features
or all specifications Intel joined. Hope you can understand.
Thanks,
Henry
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#213 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AD6CBIFSJ6G2QPPTBK743MTRLNKODANCNFSM4HY6SU4A>
.
|
Hi Oliver, I'm an Intel employee working on HAXM as well as other projects. In general, HAXM is a small project and is not intended to support all technologies Intel has. So it's possible that some features are not there. On this specific UEFI issue, do you apply the Qemu patch for HAXM ROMD support which I pasted in another issue discussion thread? From our validation, it can support OVMF boot up. Regards, |
I am going to install the patch this morning.
Regards,
Oliver Bailey
…On Tue, Apr 7, 2020, 10:47 PM hyuan3 ***@***.***> wrote:
Hi Oliver,
I'm an Intel employee working on HAXM as well as other projects. In
general, HAXM is a small project and is not intended to support all
technologies Intel has. So it's possible that some features are not there.
On this specific UEFI issue, do you apply the Qemu patch for HAXM ROMD
support which I pasted in another issue discussion thread? From our
validation, it can support OVMF boot up.
Regards,
Henry
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#213 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AD6CBIBDZRYHESBYH622HSLRLPXT7ANCNFSM4HY6SU4A>
.
|
ROM device is read-only. Write operation will generte EPT violation
and then will be handled as mmio write.
Signed-off-by: Hang Yuan hang.yuan@intel.com