[Qemu-devel] [PATCH 1/2] ivshmem: Fix compilation (wrong format specifier)

[Qemu-devel] [PATCH 1/2] ivshmem: Fix compilation (wrong format specifier)

[Stefan Weil]

st_size is an off32_t or off64_t, so %ld does not always work.

Cc: Anthony Liguori <aliguori@us.ibm.com>
Cc: Cam Macdonell <cam@cs.ualberta.ca>
Signed-off-by: Stefan Weil <weil@mail.berlios.de>
---
 hw/ivshmem.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/hw/ivshmem.c b/hw/ivshmem.c
index bbb5cba..ec894e9 100644
--- a/hw/ivshmem.c
+++ b/hw/ivshmem.c
@@ -352,8 +352,8 @@ static int check_shm_size(IVShmemState *s, int fd) {
 
     if (s->ivshmem_size > buf.st_size) {
         fprintf(stderr, "IVSHMEM ERROR: Requested memory size greater");
-        fprintf(stderr, " than shared object size (%" PRIu64 " > %ld)\n",
-                                          s->ivshmem_size, buf.st_size);
+        fprintf(stderr, " than shared object size (%" PRIu64 " > %" PRIu64 ")\n",
+                                          s->ivshmem_size, (uint64_t)buf.st_size);
         return -1;
     } else {
         return 0;
-- 
1.7.1

[Qemu-devel] [PATCH 2/2] ivshmem: Fix compilation without kvm

[Stefan Weil]

kvm_set_ioeventfd_mmio_long is only available with CONFIG_KVM.

Cc: Anthony Liguori <aliguori@us.ibm.com>
Cc: Cam Macdonell <cam@cs.ualberta.ca>
Signed-off-by: Stefan Weil <weil@mail.berlios.de>
---
 hw/ivshmem.c |    7 ++++++-
 1 files changed, 6 insertions(+), 1 deletions(-)

diff --git a/hw/ivshmem.c b/hw/ivshmem.c
index ec894e9..63562e1 100644
--- a/hw/ivshmem.c
+++ b/hw/ivshmem.c
@@ -385,8 +385,10 @@ static void close_guest_eventfds(IVShmemState *s, int posn)
     guest_curr_max = s->peers[posn].nb_eventfds;
 
     for (i = 0; i < guest_curr_max; i++) {
+#if defined(CONFIG_KVM)
         kvm_set_ioeventfd_mmio_long(s->peers[posn].eventfds[i],
                     s->mmio_addr + DOORBELL, (posn << 16) | i, 0);
+#endif
         close(s->peers[posn].eventfds[i]);
     }
 
@@ -395,7 +397,7 @@ static void close_guest_eventfds(IVShmemState *s, int posn)
 }
 
 static void setup_ioeventfds(IVShmemState *s) {
-
+#if defined(CONFIG_KVM)
     int i, j;
 
     for (i = 0; i <= s->max_peer; i++) {
@@ -404,6 +406,7 @@ static void setup_ioeventfds(IVShmemState *s) {
                     s->mmio_addr + DOORBELL, (i << 16) | j, 1);
         }
     }
+#endif
 }
 
 /* this function increase the dynamic storage need to store data about other
@@ -530,12 +533,14 @@ static void ivshmem_read(void *opaque, const uint8_t * buf, int flags)
                    guest_max_eventfd);
     }
 
+#if defined(CONFIG_KVM)
     if (ivshmem_has_feature(s, IVSHMEM_IOEVENTFD)) {
         if (kvm_set_ioeventfd_mmio_long(incoming_fd, s->mmio_addr + DOORBELL,
                         (incoming_posn << 16) | guest_max_eventfd, 1) < 0) {
             fprintf(stderr, "ivshmem: ioeventfd not available\n");
         }
     }
+#endif
 
     return;
 }
-- 
1.7.1

[Qemu-devel] Re: [PATCH 2/2] ivshmem: Fix compilation without kvm

[Paolo Bonzini]

On 08/11/2010 03:38 AM, Stefan Weil wrote:
> kvm_set_ioeventfd_mmio_long is only available with CONFIG_KVM.

We should just disable ivshmem for non-KVM

diff --git a/Makefile.target b/Makefile.target
index b791492..c8281e9 100644
--- a/Makefile.target
+++ b/Makefile.target
@@ -191,7 +191,7 @@ obj-y += rtl8139.o
  obj-y += e1000.o

  # Inter-VM PCI shared memory
-obj-y += ivshmem.o
+obj-$(CONFIG_KVM) += ivshmem.o

  # Hardware support
  obj-i386-y += vga.o

because it is also breaking Windows builds.

Alternatively, the right way to do what this patch does, is to add 
kvm_set_ioeventfd_mmio_long to kvm-stub.c, and to use 
"obj-$(CONFIG_POSIX) += ivshmem.o" in the makefile to work around the 
Windows build problems.

Paolo

Re: [Qemu-devel] Re: [PATCH 2/2] ivshmem: Fix compilation without kvm

[Blue Swirl]

[-- Attachment #1: Type: text/plain, Size: 909 bytes --]

On Wed, Aug 11, 2010 at 5:05 PM, Paolo Bonzini <pbonzini@redhat.com> wrote:
> On 08/11/2010 03:38 AM, Stefan Weil wrote:
>>
>> kvm_set_ioeventfd_mmio_long is only available with CONFIG_KVM.
>
> We should just disable ivshmem for non-KVM
>
> diff --git a/Makefile.target b/Makefile.target
> index b791492..c8281e9 100644
> --- a/Makefile.target
> +++ b/Makefile.target
> @@ -191,7 +191,7 @@ obj-y += rtl8139.o
>  obj-y += e1000.o
>
>  # Inter-VM PCI shared memory
> -obj-y += ivshmem.o
> +obj-$(CONFIG_KVM) += ivshmem.o
>
>  # Hardware support
>  obj-i386-y += vga.o
>
> because it is also breaking Windows builds.
>
> Alternatively, the right way to do what this patch does, is to add
> kvm_set_ioeventfd_mmio_long to kvm-stub.c, and to use "obj-$(CONFIG_POSIX)
> += ivshmem.o" in the makefile to work around the Windows build problems.

This patch fixes mingw32 build for me.

[-- Attachment #2: 0001-Fix-mingw32-build.patch --]
[-- Type: application/mbox, Size: 1731 bytes --]

[Qemu-devel] Re: [PATCH 2/2] ivshmem: Fix compilation without kvm

[Paolo Bonzini]

On 08/11/2010 01:18 PM, Blue Swirl wrote:
> On Wed, Aug 11, 2010 at 5:05 PM, Paolo Bonzini<pbonzini@redhat.com>  wrote:
>> On 08/11/2010 03:38 AM, Stefan Weil wrote:
>>>
>>> kvm_set_ioeventfd_mmio_long is only available with CONFIG_KVM.
>>
>> We should just disable ivshmem for non-KVM
>>
>> diff --git a/Makefile.target b/Makefile.target
>> index b791492..c8281e9 100644
>> --- a/Makefile.target
>> +++ b/Makefile.target
>> @@ -191,7 +191,7 @@ obj-y += rtl8139.o
>>   obj-y += e1000.o
>>
>>   # Inter-VM PCI shared memory
>> -obj-y += ivshmem.o
>> +obj-$(CONFIG_KVM) += ivshmem.o
>>
>>   # Hardware support
>>   obj-i386-y += vga.o
>>
>> because it is also breaking Windows builds.
>>
>> Alternatively, the right way to do what this patch does, is to add
>> kvm_set_ioeventfd_mmio_long to kvm-stub.c, and to use "obj-$(CONFIG_POSIX)
>> += ivshmem.o" in the makefile to work around the Windows build problems.
>
> This patch fixes mingw32 build for me.

Looks fine.

Paolo

[Qemu-devel] Re: [PATCH 2/2] ivshmem: Fix compilation without kvm

[Cam Macdonell]

On Wed, Aug 11, 2010 at 1:05 PM, Paolo Bonzini <pbonzini@redhat.com> wrote:
> On 08/11/2010 03:38 AM, Stefan Weil wrote:
>>
>> kvm_set_ioeventfd_mmio_long is only available with CONFIG_KVM.
>
> We should just disable ivshmem for non-KVM
>
> diff --git a/Makefile.target b/Makefile.target
> index b791492..c8281e9 100644
> --- a/Makefile.target
> +++ b/Makefile.target
> @@ -191,7 +191,7 @@ obj-y += rtl8139.o
>  obj-y += e1000.o
>
>  # Inter-VM PCI shared memory
> -obj-y += ivshmem.o
> +obj-$(CONFIG_KVM) += ivshmem.o
>
>  # Hardware support
>  obj-i386-y += vga.o
>
> because it is also breaking Windows builds.
>
> Alternatively, the right way to do what this patch does, is to add
> kvm_set_ioeventfd_mmio_long to kvm-stub.c, and to use "obj-$(CONFIG_POSIX)
> += ivshmem.o" in the makefile to work around the Windows build problems.
>

I think we decided to disable it on non-KVM systems to avoid people
stumbling into atomicity issues with emulation.

> Paolo
>
>