| Submitter | ronniesahlberg@gmail.com |
|---|---|
| Date | July 12, 2012, 6:52 a.m. |
| Message ID | <1342075967-26378-2-git-send-email-ronniesahlberg@gmail.com> |
| Download | mbox | patch |
| Permalink | /patch/170583/ |
| State | New |
| Headers | show |
Comments
Il 12/07/2012 08:52, Ronnie Sahlberg ha scritto: > Improve the tests for the LBA to cover more cases, the new test looks like this > if (r->req.cmd.lba > r->req.cmd.lba + len > || r->req.cmd.lba + len > s->qdev.max_lba) { > > For the 16 byte opcodes, the lba is a uint64, so the first check is to make sure that we do not wrap. > For example if an opcode would specify the LBA:0xffffffffffffffff and LEN:2 > then lba+len would wrap to 1. > > The second part of the test is to verify that ALL requested blocks are available, not just the first one. Fixed code style for || and applied to scsi-next. Thanks! Paolo > Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com> > --- > hw/scsi-disk.c | 6 ++++-- > 1 files changed, 4 insertions(+), 2 deletions(-) > > diff --git a/hw/scsi-disk.c b/hw/scsi-disk.c > index b2f3c0c..40c05de 100644 > --- a/hw/scsi-disk.c > +++ b/hw/scsi-disk.c > @@ -1558,7 +1558,8 @@ static int32_t scsi_send_command(SCSIRequest *req, uint8_t *buf) > if (r->req.cmd.buf[1] & 0xe0) { > goto fail; > } > - if (r->req.cmd.lba > s->qdev.max_lba) { > + if (r->req.cmd.lba > r->req.cmd.lba + len > + || r->req.cmd.lba + len > s->qdev.max_lba) { > goto illegal_lba; > } > r->sector = r->req.cmd.lba * (s->qdev.blocksize / 512); > @@ -1581,7 +1582,8 @@ static int32_t scsi_send_command(SCSIRequest *req, uint8_t *buf) > if (r->req.cmd.buf[1] & 0xe0) { > goto fail; > } > - if (r->req.cmd.lba > s->qdev.max_lba) { > + if (r->req.cmd.lba > r->req.cmd.lba + len > + || r->req.cmd.lba + len > s->qdev.max_lba) { > goto illegal_lba; > } > r->sector = r->req.cmd.lba * (s->qdev.blocksize / 512); >
Sorry, there is a bug in it.
Can you change it to :
|| r->req.cmd.lba + len > s->qdev.max_lba + 1) {
or else the last lba of the device will be flagged as out of range.
On Thu, Jul 12, 2012 at 5:04 PM, Paolo Bonzini <pbonzini@redhat.com> wrote:
> Il 12/07/2012 08:52, Ronnie Sahlberg ha scritto:
>> Improve the tests for the LBA to cover more cases, the new test looks like this
>> if (r->req.cmd.lba > r->req.cmd.lba + len
>> || r->req.cmd.lba + len > s->qdev.max_lba) {
>>
>> For the 16 byte opcodes, the lba is a uint64, so the first check is to make sure that we do not wrap.
>> For example if an opcode would specify the LBA:0xffffffffffffffff and LEN:2
>> then lba+len would wrap to 1.
>>
>> The second part of the test is to verify that ALL requested blocks are available, not just the first one.
>
> Fixed code style for || and applied to scsi-next.
>
> Thanks!
>
> Paolo
>
>> Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
>> ---
>> hw/scsi-disk.c | 6 ++++--
>> 1 files changed, 4 insertions(+), 2 deletions(-)
>>
>> diff --git a/hw/scsi-disk.c b/hw/scsi-disk.c
>> index b2f3c0c..40c05de 100644
>> --- a/hw/scsi-disk.c
>> +++ b/hw/scsi-disk.c
>> @@ -1558,7 +1558,8 @@ static int32_t scsi_send_command(SCSIRequest *req, uint8_t *buf)
>> if (r->req.cmd.buf[1] & 0xe0) {
>> goto fail;
>> }
>> - if (r->req.cmd.lba > s->qdev.max_lba) {
>> + if (r->req.cmd.lba > r->req.cmd.lba + len
>> + || r->req.cmd.lba + len > s->qdev.max_lba) {
>> goto illegal_lba;
>> }
>> r->sector = r->req.cmd.lba * (s->qdev.blocksize / 512);
>> @@ -1581,7 +1582,8 @@ static int32_t scsi_send_command(SCSIRequest *req, uint8_t *buf)
>> if (r->req.cmd.buf[1] & 0xe0) {
>> goto fail;
>> }
>> - if (r->req.cmd.lba > s->qdev.max_lba) {
>> + if (r->req.cmd.lba > r->req.cmd.lba + len
>> + || r->req.cmd.lba + len > s->qdev.max_lba) {
>> goto illegal_lba;
>> }
>> r->sector = r->req.cmd.lba * (s->qdev.blocksize / 512);
>>
>
>
Il 12/07/2012 09:06, ronnie sahlberg ha scritto: > Sorry, there is a bug in it. > > Can you change it to : > || r->req.cmd.lba + len > s->qdev.max_lba + 1) { > > or else the last lba of the device will be flagged as out of range. Yes, done that already. Can you add a test for the last LBA to your testsuite? Paolo
Patch
diff --git a/hw/scsi-disk.c b/hw/scsi-disk.c index b2f3c0c..40c05de 100644 --- a/hw/scsi-disk.c +++ b/hw/scsi-disk.c @@ -1558,7 +1558,8 @@ static int32_t scsi_send_command(SCSIRequest *req, uint8_t *buf) if (r->req.cmd.buf[1] & 0xe0) { goto fail; } - if (r->req.cmd.lba > s->qdev.max_lba) { + if (r->req.cmd.lba > r->req.cmd.lba + len + || r->req.cmd.lba + len > s->qdev.max_lba) { goto illegal_lba; } r->sector = r->req.cmd.lba * (s->qdev.blocksize / 512); @@ -1581,7 +1582,8 @@ static int32_t scsi_send_command(SCSIRequest *req, uint8_t *buf) if (r->req.cmd.buf[1] & 0xe0) { goto fail; } - if (r->req.cmd.lba > s->qdev.max_lba) { + if (r->req.cmd.lba > r->req.cmd.lba + len + || r->req.cmd.lba + len > s->qdev.max_lba) { goto illegal_lba; } r->sector = r->req.cmd.lba * (s->qdev.blocksize / 512);
Improve the tests for the LBA to cover more cases, the new test looks like this if (r->req.cmd.lba > r->req.cmd.lba + len || r->req.cmd.lba + len > s->qdev.max_lba) { For the 16 byte opcodes, the lba is a uint64, so the first check is to make sure that we do not wrap. For example if an opcode would specify the LBA:0xffffffffffffffff and LEN:2 then lba+len would wrap to 1. The second part of the test is to verify that ALL requested blocks are available, not just the first one. Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com> --- hw/scsi-disk.c | 6 ++++-- 1 files changed, 4 insertions(+), 2 deletions(-)