2020-11-10 22:14:55

by Colin King

[permalink] [raw]
Subject: [PATCH][V2] PCI: Fix a potential uninitentional integer overflow issue

From: Colin Ian King <[email protected]>

The shift of 1 by align_order is evaluated using 32 bit arithmetic
and the result is assigned to a resource_size_t type variable that
is a 64 bit unsigned integer on 64 bit platforms. Fix an overflow
before widening issue by making the 1 a ULL.

Addresses-Coverity: ("Unintentional integer overflow")
Fixes: 07d8d7e57c28 ("PCI: Make specifying PCI devices in kernel parameters reusable")
Signed-off-by: Colin Ian King <[email protected]>
---

V2: Use ULL instead of BIT_ULL(), fix spelling mistake and capitalize first
word of patch subject.

---
drivers/pci/pci.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index 3ef63a101fa1..248044a7ef8c 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -6214,7 +6214,7 @@ static resource_size_t pci_specified_resource_alignment(struct pci_dev *dev,
if (align_order == -1)
align = PAGE_SIZE;
else
- align = 1 << align_order;
+ align = 1ULL << align_order;
break;
} else if (ret < 0) {
pr_err("PCI: Can't parse resource_alignment parameter: %s\n",
--
2.28.0


2020-11-10 23:59:07

by Logan Gunthorpe

[permalink] [raw]
Subject: Re: [PATCH][V2] PCI: Fix a potential uninitentional integer overflow issue



On 2020-11-10 3:10 p.m., Colin King wrote:
> From: Colin Ian King <[email protected]>
>
> The shift of 1 by align_order is evaluated using 32 bit arithmetic
> and the result is assigned to a resource_size_t type variable that
> is a 64 bit unsigned integer on 64 bit platforms. Fix an overflow
> before widening issue by making the 1 a ULL.
>
> Addresses-Coverity: ("Unintentional integer overflow")
> Fixes: 07d8d7e57c28 ("PCI: Make specifying PCI devices in kernel parameters reusable")

I think this should probably be

Fixes: 32a9a682bef2 ("PCI: allow assignment of memory resources with a
specified alignment")

That is the commit where the original bug was introduced. 644a544fd9bcd
then extends the code a little bit and 07d8d7e57c28 only refactors it
into a reusable function. If we want this in older stable kernels then
we will probably need to make different patches for the other two vintages.

> Signed-off-by: Colin Ian King <[email protected]>

Besides that, the change makes sense to me.

Reviewed-by: Logan Gunthorpe <[email protected]>

Thanks,

Logan

2020-11-14 22:02:28

by Bjorn Helgaas

[permalink] [raw]
Subject: Re: [PATCH][V2] PCI: Fix a potential uninitentional integer overflow issue

[+cc Dan]

On Tue, Nov 10, 2020 at 10:10:48PM +0000, Colin King wrote:
> From: Colin Ian King <[email protected]>
>
> The shift of 1 by align_order is evaluated using 32 bit arithmetic
> and the result is assigned to a resource_size_t type variable that
> is a 64 bit unsigned integer on 64 bit platforms. Fix an overflow
> before widening issue by making the 1 a ULL.
>
> Addresses-Coverity: ("Unintentional integer overflow")
> Fixes: 07d8d7e57c28 ("PCI: Make specifying PCI devices in kernel parameters reusable")
> Signed-off-by: Colin Ian King <[email protected]>

Applied to pci/misc for v5.11 with Logan's Reviewed-by and also the
Fixes: correction.

I first applied the patch below to bounds-check the alignment as noted
by Dan.

> ---
>
> V2: Use ULL instead of BIT_ULL(), fix spelling mistake and capitalize first
> word of patch subject.
>
> ---
> drivers/pci/pci.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index 3ef63a101fa1..248044a7ef8c 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -6214,7 +6214,7 @@ static resource_size_t pci_specified_resource_alignment(struct pci_dev *dev,
> if (align_order == -1)
> align = PAGE_SIZE;
> else
> - align = 1 << align_order;
> + align = 1ULL << align_order;
> break;
> } else if (ret < 0) {
> pr_err("PCI: Can't parse resource_alignment parameter: %s\n",

commit d6ca242c448f ("PCI: Bounds-check command-line resource alignment requests")
Author: Bjorn Helgaas <[email protected]>
Date: Thu Nov 5 14:51:36 2020 -0600

PCI: Bounds-check command-line resource alignment requests

32-bit BARs are limited to 2GB size (2^31). By extension, I assume 64-bit
BARs are limited to 2^63 bytes. Limit the alignment requested by the
"pci=resource_alignment=" command-line parameter to 2^63.

Link: https://lore.kernel.org/r/20201007123045.GS4282@kadam
Reported-by: Dan Carpenter <[email protected]>
Signed-off-by: Bjorn Helgaas <[email protected]>

diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index 8b9bea8ba751..26c1b2d0bacd 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -6197,19 +6197,21 @@ static resource_size_t pci_specified_resource_alignment(struct pci_dev *dev,
while (*p) {
count = 0;
if (sscanf(p, "%d%n", &align_order, &count) == 1 &&
- p[count] == '@') {
+ p[count] == '@') {
p += count + 1;
+ if (align_order > 63) {
+ pr_err("PCI: Invalid requested alignment (order %d)\n",
+ align_order);
+ align_order = PAGE_SHIFT;
+ }
} else {
- align_order = -1;
+ align_order = PAGE_SHIFT;
}

ret = pci_dev_str_match(dev, p, &p);
if (ret == 1) {
*resize = true;
- if (align_order == -1)
- align = PAGE_SIZE;
- else
- align = 1 << align_order;
+ align = 1 << align_order;
break;
} else if (ret < 0) {
pr_err("PCI: Can't parse resource_alignment parameter: %s\n",

2020-11-14 22:12:51

by Colin King

[permalink] [raw]
Subject: Re: [PATCH][V2] PCI: Fix a potential uninitentional integer overflow issue

On 14/11/2020 21:53, Bjorn Helgaas wrote:
> [+cc Dan]
>
> On Tue, Nov 10, 2020 at 10:10:48PM +0000, Colin King wrote:
>> From: Colin Ian King <[email protected]>
>>
>> The shift of 1 by align_order is evaluated using 32 bit arithmetic
>> and the result is assigned to a resource_size_t type variable that
>> is a 64 bit unsigned integer on 64 bit platforms. Fix an overflow
>> before widening issue by making the 1 a ULL.
>>
>> Addresses-Coverity: ("Unintentional integer overflow")
>> Fixes: 07d8d7e57c28 ("PCI: Make specifying PCI devices in kernel parameters reusable")
>> Signed-off-by: Colin Ian King <[email protected]>
>
> Applied to pci/misc for v5.11 with Logan's Reviewed-by and also the
> Fixes: correction.
>
> I first applied the patch below to bounds-check the alignment as noted
> by Dan.
>
>> ---
>>
>> V2: Use ULL instead of BIT_ULL(), fix spelling mistake and capitalize first
>> word of patch subject.
>>
>> ---
>> drivers/pci/pci.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
>> index 3ef63a101fa1..248044a7ef8c 100644
>> --- a/drivers/pci/pci.c
>> +++ b/drivers/pci/pci.c
>> @@ -6214,7 +6214,7 @@ static resource_size_t pci_specified_resource_alignment(struct pci_dev *dev,
>> if (align_order == -1)
>> align = PAGE_SIZE;
>> else
>> - align = 1 << align_order;
>> + align = 1ULL << align_order;
>> break;
>> } else if (ret < 0) {
>> pr_err("PCI: Can't parse resource_alignment parameter: %s\n",
>
> commit d6ca242c448f ("PCI: Bounds-check command-line resource alignment requests")
> Author: Bjorn Helgaas <[email protected]>
> Date: Thu Nov 5 14:51:36 2020 -0600
>
> PCI: Bounds-check command-line resource alignment requests
>
> 32-bit BARs are limited to 2GB size (2^31). By extension, I assume 64-bit
> BARs are limited to 2^63 bytes. Limit the alignment requested by the
> "pci=resource_alignment=" command-line parameter to 2^63.
>
> Link: https://lore.kernel.org/r/20201007123045.GS4282@kadam
> Reported-by: Dan Carpenter <[email protected]>
> Signed-off-by: Bjorn Helgaas <[email protected]>
>
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index 8b9bea8ba751..26c1b2d0bacd 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -6197,19 +6197,21 @@ static resource_size_t pci_specified_resource_alignment(struct pci_dev *dev,
> while (*p) {
> count = 0;
> if (sscanf(p, "%d%n", &align_order, &count) == 1 &&
> - p[count] == '@') {
> + p[count] == '@') {
> p += count + 1;
> + if (align_order > 63) {
> + pr_err("PCI: Invalid requested alignment (order %d)\n",
> + align_order);
> + align_order = PAGE_SHIFT;
> + }
> } else {
> - align_order = -1;
> + align_order = PAGE_SHIFT;
> }
>
> ret = pci_dev_str_match(dev, p, &p);
> if (ret == 1) {
> *resize = true;
> - if (align_order == -1)
> - align = PAGE_SIZE;
> - else
> - align = 1 << align_order;
> + align = 1 << align_order;
> break;
> } else if (ret < 0) {
> pr_err("PCI: Can't parse resource_alignment parameter: %s\n",
>

Thanks.