* [PATCH] fix extboot from boot with cache=off
@ 2009-03-17 2:52 Nolan
2009-03-19 10:03 ` Avi Kivity
0 siblings, 1 reply; 4+ messages in thread
From: Nolan @ 2009-03-17 2:52 UTC (permalink / raw)
To: kvm
Extboot submits requests with whatever buffer alignment the guest gave
to the BIOS. This breaks with O_DIRECT disks, as they require 512 byte
alignment.
Most guest bootloaders sector align their requests out of paranoia, but
the OpenBSD bootloader does not.
This patch always copies. Since extboot is only used at boot time to
load limited amounts of data, the overhead is not problematic. I also
switched to using cpu_physical_memory_* instead of groveling around with
phys_ram_base directly.
Signed-off-by: Nolan Leake <nolan <at> sigbus.net>
---
qemu/hw/extboot.c | 35 ++++++++++++++++++++++++-----------
1 files changed, 24 insertions(+), 11 deletions(-)
diff --git a/qemu/hw/extboot.c b/qemu/hw/extboot.c
index ada0fdd..ea16b3e 100644
--- a/qemu/hw/extboot.c
+++ b/qemu/hw/extboot.c
@@ -77,19 +77,29 @@ static void extboot_write_cmd(void *opaque, uint32_t addr, uint32_t value)
BlockDriverState *bs = opaque;
int cylinders, heads, sectors, err;
uint64_t nb_sectors;
-
- get_translated_chs(bs, &cylinders, &heads, §ors);
+ target_phys_addr_t pa;
+ int blen;
+ void *buf = NULL;
if (cmd->type == 0x01 || cmd->type == 0x02) {
- target_ulong pa = cmd->xfer.segment * 16 + cmd->xfer.segment;
+ pa = cmd->xfer.segment * 16 + cmd->xfer.offset;
+ blen = cmd->xfer.nb_sectors * 512;
+ buf = qemu_memalign(512, blen);
+ if (!buf) {
+ printf("qemu_memalign failed\n");
+ return;
+ }
/* possible buffer overflow */
- if ((pa + cmd->xfer.nb_sectors * 512) > phys_ram_size)
+ if ((pa + blen) > phys_ram_size) {
+ qemu_free(buf);
return;
+ }
}
switch (cmd->type) {
case 0x00:
+ get_translated_chs(bs, &cylinders, &heads, §ors);
bdrv_get_geometry(bs, &nb_sectors);
cmd->query_geometry.cylinders = cylinders;
cmd->query_geometry.heads = heads;
@@ -98,22 +108,25 @@ static void extboot_write_cmd(void *opaque, uint32_t addr, uint32_t value)
cpu_physical_memory_set_dirty((value & 0xFFFF) << 4);
break;
case 0x01:
- err = bdrv_read(bs, cmd->xfer.sector, phys_ram_base +
- cmd->xfer.segment * 16 + cmd->xfer.offset,
- cmd->xfer.nb_sectors);
+ err = bdrv_read(bs, cmd->xfer.sector, buf, cmd->xfer.nb_sectors);
if (err)
printf("Read failed\n");
+
+ cpu_physical_memory_write(pa, buf, blen);
+
break;
case 0x02:
- err = bdrv_write(bs, cmd->xfer.sector, phys_ram_base +
- cmd->xfer.segment * 16 + cmd->xfer.offset,
- cmd->xfer.nb_sectors);
+ cpu_physical_memory_read(pa, buf, blen);
+
+ err = bdrv_write(bs, cmd->xfer.sector, buf, cmd->xfer.nb_sectors);
if (err)
printf("Write failed\n");
- cpu_physical_memory_set_dirty(cmd->xfer.segment * 16 + cmd->xfer.offset);
break;
}
+
+ if (buf)
+ qemu_free(buf);
}
void extboot_init(BlockDriverState *bs, int cmd)
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH] fix extboot from boot with cache=off
2009-03-17 2:52 [PATCH] fix extboot from boot with cache=off Nolan
@ 2009-03-19 10:03 ` Avi Kivity
2009-03-20 0:02 ` Nolan
0 siblings, 1 reply; 4+ messages in thread
From: Avi Kivity @ 2009-03-19 10:03 UTC (permalink / raw)
To: Nolan; +Cc: kvm
Nolan wrote:
> Extboot submits requests with whatever buffer alignment the guest gave
> to the BIOS. This breaks with O_DIRECT disks, as they require 512 byte
> alignment.
>
> Most guest bootloaders sector align their requests out of paranoia, but
> the OpenBSD bootloader does not.
>
> This patch always copies. Since extboot is only used at boot time to
> load limited amounts of data, the overhead is not problematic. I also
> switched to using cpu_physical_memory_* instead of groveling around with
> phys_ram_base directly.
>
> Signed-off-by: Nolan Leake <nolan <at> sigbus.net>
>
> ---
>
> qemu/hw/extboot.c | 35 ++++++++++++++++++++++++-----------
> 1 files changed, 24 insertions(+), 11 deletions(-)
>
> diff --git a/qemu/hw/extboot.c b/qemu/hw/extboot.c
> index ada0fdd..ea16b3e 100644
> --- a/qemu/hw/extboot.c
> +++ b/qemu/hw/extboot.c
> @@ -77,19 +77,29 @@ static void extboot_write_cmd(void *opaque, uint32_t addr, uint32_t value)
> BlockDriverState *bs = opaque;
> int cylinders, heads, sectors, err;
> uint64_t nb_sectors;
> -
> - get_translated_chs(bs, &cylinders, &heads, §ors);
> + target_phys_addr_t pa;
> + int blen;
> + void *buf = NULL;
>
> if (cmd->type == 0x01 || cmd->type == 0x02) {
> - target_ulong pa = cmd->xfer.segment * 16 + cmd->xfer.segment;
> + pa = cmd->xfer.segment * 16 + cmd->xfer.offset;
> + blen = cmd->xfer.nb_sectors * 512;
> + buf = qemu_memalign(512, blen);
> + if (!buf) {
> + printf("qemu_memalign failed\n");
> + return;
> + }
>
Don't check for allocation failures. qemu_malloc() terminates
gracefully on failure, qemu_memalign() does not, but should.
>
> /* possible buffer overflow */
> - if ((pa + cmd->xfer.nb_sectors * 512) > phys_ram_size)
> + if ((pa + blen) > phys_ram_size) {
> + qemu_free(buf);
> return;
> + }
>
cpu_physical_memory_rw() will check these for you.
All in all, a nice improvement in addition to the bug fix.
--
error compiling committee.c: too many arguments to function
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] fix extboot from boot with cache=off
2009-03-19 10:03 ` Avi Kivity
@ 2009-03-20 0:02 ` Nolan
2009-03-22 9:14 ` Avi Kivity
0 siblings, 1 reply; 4+ messages in thread
From: Nolan @ 2009-03-20 0:02 UTC (permalink / raw)
To: Avi Kivity; +Cc: kvm
Thanks for the prompt review!
On Thu, 2009-03-19 at 12:03 +0200, Avi Kivity wrote:
> Nolan wrote:
> > if (cmd->type == 0x01 || cmd->type == 0x02) {
> > - target_ulong pa = cmd->xfer.segment * 16 + cmd->xfer.segment;
> > + pa = cmd->xfer.segment * 16 + cmd->xfer.offset;
> > + blen = cmd->xfer.nb_sectors * 512;
> > + buf = qemu_memalign(512, blen);
> > + if (!buf) {
> > + printf("qemu_memalign failed\n");
> > + return;
> > + }
>
> Don't check for allocation failures. qemu_malloc() terminates
> gracefully on failure, qemu_memalign() does not, but should.
OK, I've removed the check. As long as I'm in here, I've removed the
check of the return value of qemu_mallocz in extboot_init as well.
I'll send a separate patch to qemu-devel with the change to make
qemu_memalign abort on failure.
> > /* possible buffer overflow */
> > - if ((pa + cmd->xfer.nb_sectors * 512) > phys_ram_size)
> > + if ((pa + blen) > phys_ram_size) {
> > + qemu_free(buf);
> > return;
> > + }
> >
>
> cpu_physical_memory_rw() will check these for you.
Check removed.
New patch appended:
Signed-off-by: Nolan Leake <nolan <at> sigbus.net>
---
qemu/hw/extboot.c | 35 +++++++++++++++++------------------
1 files changed, 17 insertions(+), 18 deletions(-)
diff --git a/qemu/hw/extboot.c b/qemu/hw/extboot.c
index ada0fdd..32e6226 100644
--- a/qemu/hw/extboot.c
+++ b/qemu/hw/extboot.c
@@ -77,19 +77,19 @@ static void extboot_write_cmd(void *opaque, uint32_t addr, uint32_t value)
BlockDriverState *bs = opaque;
int cylinders, heads, sectors, err;
uint64_t nb_sectors;
-
- get_translated_chs(bs, &cylinders, &heads, §ors);
+ target_phys_addr_t pa;
+ int blen;
+ void *buf = NULL;
if (cmd->type == 0x01 || cmd->type == 0x02) {
- target_ulong pa = cmd->xfer.segment * 16 + cmd->xfer.segment;
-
- /* possible buffer overflow */
- if ((pa + cmd->xfer.nb_sectors * 512) > phys_ram_size)
- return;
+ pa = cmd->xfer.segment * 16 + cmd->xfer.offset;
+ blen = cmd->xfer.nb_sectors * 512;
+ buf = qemu_memalign(512, blen);
}
switch (cmd->type) {
case 0x00:
+ get_translated_chs(bs, &cylinders, &heads, §ors);
bdrv_get_geometry(bs, &nb_sectors);
cmd->query_geometry.cylinders = cylinders;
cmd->query_geometry.heads = heads;
@@ -98,22 +98,25 @@ static void extboot_write_cmd(void *opaque, uint32_t addr, uint32_t value)
cpu_physical_memory_set_dirty((value & 0xFFFF) << 4);
break;
case 0x01:
- err = bdrv_read(bs, cmd->xfer.sector, phys_ram_base +
- cmd->xfer.segment * 16 + cmd->xfer.offset,
- cmd->xfer.nb_sectors);
+ err = bdrv_read(bs, cmd->xfer.sector, buf, cmd->xfer.nb_sectors);
if (err)
printf("Read failed\n");
+
+ cpu_physical_memory_write(pa, buf, blen);
+
break;
case 0x02:
- err = bdrv_write(bs, cmd->xfer.sector, phys_ram_base +
- cmd->xfer.segment * 16 + cmd->xfer.offset,
- cmd->xfer.nb_sectors);
+ cpu_physical_memory_read(pa, buf, blen);
+
+ err = bdrv_write(bs, cmd->xfer.sector, buf, cmd->xfer.nb_sectors);
if (err)
printf("Write failed\n");
- cpu_physical_memory_set_dirty(cmd->xfer.segment * 16 + cmd->xfer.offset);
break;
}
+
+ if (buf)
+ qemu_free(buf);
}
void extboot_init(BlockDriverState *bs, int cmd)
@@ -121,10 +124,6 @@ void extboot_init(BlockDriverState *bs, int cmd)
int *pcmd;
pcmd = qemu_mallocz(sizeof(int));
- if (!pcmd) {
- fprintf(stderr, "Error allocating memory\n");
- exit(1);
- }
*pcmd = cmd;
register_ioport_read(0x404, 1, 1, extboot_read, pcmd);
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH] fix extboot from boot with cache=off
2009-03-20 0:02 ` Nolan
@ 2009-03-22 9:14 ` Avi Kivity
0 siblings, 0 replies; 4+ messages in thread
From: Avi Kivity @ 2009-03-22 9:14 UTC (permalink / raw)
To: Nolan; +Cc: kvm
Nolan wrote:
> New patch appended:
>
>
Applied, thanks.
--
error compiling committee.c: too many arguments to function
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2009-03-22 9:14 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2009-03-17 2:52 [PATCH] fix extboot from boot with cache=off Nolan
2009-03-19 10:03 ` Avi Kivity
2009-03-20 0:02 ` Nolan
2009-03-22 9:14 ` Avi Kivity
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox