* Re: [Qemu-devel] [PATCH] bugfix: passing reference instead of value
2015-12-28 11:50 ` Michael S. Tsirkin
@ 2015-12-28 12:30 ` Cao jin
2015-12-28 12:39 ` Cao jin
1 sibling, 0 replies; 4+ messages in thread
From: Cao jin @ 2015-12-28 12:30 UTC (permalink / raw)
To: Michael S. Tsirkin; +Cc: qemu-devel, stefano.stabellini
On 12/28/2015 07:50 PM, Michael S. Tsirkin wrote:
> On Mon, Dec 28, 2015 at 10:54:23AM +0800, Cao jin wrote:
>> Fix the bug introduced by 595a4f07. Function host_pci_config_read() should be
>> passed by a reference, not a value, for the later pci_default_write_config().
>
> What's the effect of the bug? Does it break igd assignment?
> How come it worked for people?
> If the function is never called, mayber we can get rid
> of it completely?
>
sorry if I didn`t explain it clearly to you. let me try the explanation
again: This function is called only when using
TYPE_IGD_PASSTHROUGH_I440FX_PCI_DEVICE(when realize it)
the effect of the bug:
pci_default_write_config(pci_dev, pos, val, len);
^ *its value is always 0*
I think it won`t break igd assignment, but just give a wrong register
value(forever 0) in PCI config space(wrong register value may results in
abnormal working state?). the register should get its value by
host_pci_config_read()
Because my bad English description, Let me do a analogy, here is the
imitation of original code:
void swap(unsigned int val) //this is host_pci_config_read()
{
unsigned int org = 2;
memcpy(&val, &org, sizeof(unsigned int));
}
int main()
{
unsigned int val = 0;
swap(val);
printf("val = %d\n", val);
return 0;
}
author want to get: val = 2. but it will always: val = 0; This is
exactly the bug I find.
> Stefano?
>
>>
>> Signed-off-by: Cao jin <caoj.fnst@cn.fujitsu.com>
>> ---
>> Separated from previous "igd-passthru convert to realize" patch. Since these
>> two don`t have dependency, can send it solely.
>>
>> Not test since it is easy to find out if reading carefully, just compiled.
>>
>> hw/pci-host/piix.c | 8 +++++---
>> 1 file changed, 5 insertions(+), 3 deletions(-)
>>
>> diff --git a/hw/pci-host/piix.c b/hw/pci-host/piix.c
>> index 715208b..924f0fa 100644
>> --- a/hw/pci-host/piix.c
>> +++ b/hw/pci-host/piix.c
>> @@ -761,7 +761,7 @@ static const IGDHostInfo igd_host_bridge_infos[] = {
>> {0xa8, 4}, /* SNB: base of GTT stolen memory */
>> };
>>
>> -static int host_pci_config_read(int pos, int len, uint32_t val)
>> +static int host_pci_config_read(int pos, int len, uint32_t *val)
>> {
>> char path[PATH_MAX];
>> int config_fd;
>> @@ -784,12 +784,14 @@ static int host_pci_config_read(int pos, int len, uint32_t val)
>> ret = -errno;
>> goto out;
>> }
>> +
>> do {
>> - rc = read(config_fd, (uint8_t *)&val, len);
>> + rc = read(config_fd, (uint8_t *)val, len);
>> } while (rc < 0 && (errno == EINTR || errno == EAGAIN));
>> if (rc != len) {
>> ret = -errno;
>> }
>> +
>> out:
>> close(config_fd);
>> return ret;
>> @@ -805,7 +807,7 @@ static int igd_pt_i440fx_initfn(struct PCIDevice *pci_dev)
>> for (i = 0; i < num; i++) {
>> pos = igd_host_bridge_infos[i].offset;
>> len = igd_host_bridge_infos[i].len;
>> - rc = host_pci_config_read(pos, len, val);
>> + rc = host_pci_config_read(pos, len, &val);
>> if (rc) {
>> return -ENODEV;
>> }
>> --
>> 2.1.0
>>
>>
>
>
> .
>
--
Yours Sincerely,
Cao Jin
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [Qemu-devel] [PATCH] bugfix: passing reference instead of value
2015-12-28 11:50 ` Michael S. Tsirkin
2015-12-28 12:30 ` Cao jin
@ 2015-12-28 12:39 ` Cao jin
1 sibling, 0 replies; 4+ messages in thread
From: Cao jin @ 2015-12-28 12:39 UTC (permalink / raw)
To: Michael S. Tsirkin; +Cc: qemu-devel, stefano.stabellini
BTW, I send the v2 version of this patch, the changelog of v2 is:
ensure the value writen into register of pci config space is always
little endian, using cpu_to_le32().
So, the actual change by v2 is following:
- pci_default_write_config(pci_dev, pos, val, len);
+ pci_default_write_config(pci_dev, pos, cpu_to_le32(val), len);
Maybe we can discuss in v2 thread.
On 12/28/2015 07:50 PM, Michael S. Tsirkin wrote:
> On Mon, Dec 28, 2015 at 10:54:23AM +0800, Cao jin wrote:
>> Fix the bug introduced by 595a4f07. Function host_pci_config_read() should be
>> passed by a reference, not a value, for the later pci_default_write_config().
>
> What's the effect of the bug? Does it break igd assignment?
> How come it worked for people?
> If the function is never called, mayber we can get rid
> of it completely?
>
> Stefano?
>
>>
>> Signed-off-by: Cao jin <caoj.fnst@cn.fujitsu.com>
>> ---
>> Separated from previous "igd-passthru convert to realize" patch. Since these
>> two don`t have dependency, can send it solely.
>>
>> Not test since it is easy to find out if reading carefully, just compiled.
>>
>> hw/pci-host/piix.c | 8 +++++---
>> 1 file changed, 5 insertions(+), 3 deletions(-)
>>
>> diff --git a/hw/pci-host/piix.c b/hw/pci-host/piix.c
>> index 715208b..924f0fa 100644
>> --- a/hw/pci-host/piix.c
>> +++ b/hw/pci-host/piix.c
>> @@ -761,7 +761,7 @@ static const IGDHostInfo igd_host_bridge_infos[] = {
>> {0xa8, 4}, /* SNB: base of GTT stolen memory */
>> };
>>
>> -static int host_pci_config_read(int pos, int len, uint32_t val)
>> +static int host_pci_config_read(int pos, int len, uint32_t *val)
>> {
>> char path[PATH_MAX];
>> int config_fd;
>> @@ -784,12 +784,14 @@ static int host_pci_config_read(int pos, int len, uint32_t val)
>> ret = -errno;
>> goto out;
>> }
>> +
>> do {
>> - rc = read(config_fd, (uint8_t *)&val, len);
>> + rc = read(config_fd, (uint8_t *)val, len);
>> } while (rc < 0 && (errno == EINTR || errno == EAGAIN));
>> if (rc != len) {
>> ret = -errno;
>> }
>> +
>> out:
>> close(config_fd);
>> return ret;
>> @@ -805,7 +807,7 @@ static int igd_pt_i440fx_initfn(struct PCIDevice *pci_dev)
>> for (i = 0; i < num; i++) {
>> pos = igd_host_bridge_infos[i].offset;
>> len = igd_host_bridge_infos[i].len;
>> - rc = host_pci_config_read(pos, len, val);
>> + rc = host_pci_config_read(pos, len, &val);
>> if (rc) {
>> return -ENODEV;
>> }
>> --
>> 2.1.0
>>
>>
>
>
> .
>
--
Yours Sincerely,
Cao Jin
^ permalink raw reply [flat|nested] 4+ messages in thread