[PATCH for-rc 0/2] Fixes for module unload

[PATCH for-rc 0/2] Fixes for module unload

[Dennis Dalessandro]

Two fixes for unload path. One is where the module use count got messed up in a
previous fix, we missed hitting these paths. I missed hitting these paths. The
other patch is going back to kfree for the dummy netdev. We are currently
re-working how this dummy netdev stuff works and will send a series for-next
soon.

---

Dennis Dalessandro (2):
      IB/hfi1: Restore kfree in dummy_netdev cleanup
      IB/hfi1: Fix module use count flaw due to leftover module put calls


 drivers/infiniband/hw/hfi1/debugfs.c   |   19 ++-----------------
 drivers/infiniband/hw/hfi1/netdev_rx.c |    2 +-
 2 files changed, 3 insertions(+), 18 deletions(-)

--
-Denny

[PATCH for-rc 1/2] IB/hfi1: Restore kfree in dummy_netdev cleanup

[Dennis Dalessandro]

We need to do some rework on the dummy netdev. Calling the free_netdev() would
normally make sense, and that will be addressed in an upcoming patch. For now
just revert the behavior to what it was before keeping the unused variable
removal part of the patch.

The dd->dumm_netdev is mainly used for packet receiving through
NAPI. Consequently, it is allocated with kcalloc_node instead of
alloc_netdev_mqs for typical net devices. A a result, it should be
freed with kfree instead of free_netdev that leads to a crash when
unloading the hfi1 module:

[32200.283994] ########### Unload/Load : Unloading HFI driver on
phwtpriv58.ph.intel.com #########
[32200.455609] infiniband hfi1_0: removing VNIC client
[32200.565845] hfi1 0000:05:00.0: hfi1_0: hfi1 netdev freed
[32200.572643] BUG: kernel NULL pointer dereference, address: 0000000000000000
[32200.581028] #PF: supervisor read access in kernel mode
[32200.587330] #PF: error_code(0x0000) - not-present page
[32200.593614] PGD 8000000855b54067 P4D 8000000855b54067 PUD 84a4f5067 PMD 0
[32200.601899] Oops: 0000 [#1] SMP PTI
[32200.606346] CPU: 73 PID: 10299 Comm: modprobe Not tainted 5.6.0-rc5+ #1
[32200.614292] Hardware name: Intel Corporation S2600WT2R/S2600WT2R, BIOS
SE5C610.86B.01.01.0016.033120161139 03/31/2016
[32200.626704] RIP: 0010:__hw_addr_flush+0x12/0x80
[32200.632310] Code: 40 00 48 83 c4 08 4c 89 e7 5b 5d 41 5c e9 76 77 18 00 66 0f
1f 44 00 00 0f 1f 44 00 00 41 54 49 89 fc 55 53 48 8b 1f 48 39 df <48> 8b 2b 75
08 eb 4a 48 89 eb 48 89 c5 48 89 df e8 99 bf d0 ff 84
[32200.654441] RSP: 0018:ffffb40e08783db8 EFLAGS: 00010282
[32200.660901] RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000002
[32200.669422] RDX: ffffb40e00000000 RSI: 0000000000000246 RDI: ffff88ab13662298
[32200.677997] RBP: ffff88ab13662000 R08: 0000000000001549 R09: 0000000000001549
[32200.686508] R10: 0000000000000001 R11: 0000000000aaaaaa R12: ffff88ab13662298
[32200.695085] R13: ffff88ab1b259e20 R14: ffff88ab1b259e42 R15: 0000000000000000
[32200.703627] FS:  00007fb39b534740(0000) GS:ffff88b31f940000(0000)
knlGS:0000000000000000
[32200.713263] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[32200.720278] CR2: 0000000000000000 CR3: 000000084d3ea004 CR4: 00000000003606e0
[32200.728865] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[32200.737386] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[32200.745976] Call Trace:
[32200.749298]  dev_addr_flush+0x15/0x30
[32200.753989]  free_netdev+0x7e/0x130
[32200.758450]  hfi1_netdev_free+0x59/0x70 [hfi1]
[32200.764043]  remove_one+0x65/0x110 [hfi1]
[32200.769130]  pci_device_remove+0x3b/0xc0
[32200.774121]  device_release_driver_internal+0xec/0x1b0
[32200.780404]  driver_detach+0x46/0x90
[32200.784982]  bus_remove_driver+0x58/0xd0
[32200.789970]  pci_unregister_driver+0x26/0xa0
[32200.795309]  hfi1_mod_cleanup+0xc/0xd54 [hfi1]
[32200.800844]  __x64_sys_delete_module+0x16c/0x260
[32200.806513]  ? exit_to_usermode_loop+0xa4/0xc0
[32200.812054]  do_syscall_64+0x5b/0x200
[32200.816655]  entry_SYSCALL_64_after_hwframe+0x44/0xa9

Fixes: 3584adcd8898 ("IB/hfi1: Use free_netdev() in hfi1_netdev_free()")
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
---
 drivers/infiniband/hw/hfi1/netdev_rx.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/hw/hfi1/netdev_rx.c b/drivers/infiniband/hw/hfi1/netdev_rx.c
index 63688e8..6d263c9 100644
--- a/drivers/infiniband/hw/hfi1/netdev_rx.c
+++ b/drivers/infiniband/hw/hfi1/netdev_rx.c
@@ -373,7 +373,7 @@ void hfi1_netdev_free(struct hfi1_devdata *dd)
 {
 	if (dd->dummy_netdev) {
 		dd_dev_info(dd, "hfi1 netdev freed\n");
-		free_netdev(dd->dummy_netdev);
+		kfree(dd->dummy_netdev);
 		dd->dummy_netdev = NULL;
 	}
 }

Re: [PATCH for-rc 1/2] IB/hfi1: Restore kfree in dummy_netdev cleanup

[Jason Gunthorpe]

On Tue, Jun 23, 2020 at 04:32:24PM -0400, Dennis Dalessandro wrote:
> We need to do some rework on the dummy netdev. Calling the free_netdev() would
> normally make sense, and that will be addressed in an upcoming patch. For now
> just revert the behavior to what it was before keeping the unused variable
> removal part of the patch.
> 
> The dd->dumm_netdev is mainly used for packet receiving through
> NAPI. Consequently, it is allocated with kcalloc_node instead of
> alloc_netdev_mqs for typical net devices.

Gross - make sure your rework allocates netdevs using alloc_netdev.

Jason

RE: [PATCH for-rc 1/2] IB/hfi1: Restore kfree in dummy_netdev cleanup

[Marciniszyn, Mike]

> 
> Gross - make sure your rework allocates netdevs using alloc_netdev.
> 

I don't think that is the correct solution.

All other users of the dummy net device embed it in other structures:

init_dummy_netdev(&mal->dummy_dev);
init_dummy_netdev(&eth->dummy_dev);
init_dummy_netdev(&ar->napi_dev);
init_dummy_netdev(&irq_grp->napi_ndev);
init_dummy_netdev(&wil->napi_ndev);
init_dummy_netdev(&trans_pcie->napi_dev);
init_dummy_netdev(&dev->napi_dev);
init_dummy_netdev(&bus->mux_dev);

There is NO explicit free of the dummy since its lifetime is controlled by its parent struct.

The fault with the current AIP use of the dummy netdev is that it confuses an is-a with a has-a relationship.

This code highlights the confusion by extending the allocation to include the real rx data:

	const int netdev_size = sizeof(*dd->dummy_netdev) +
		sizeof(struct hfi1_netdev_priv);
	<snip>
	dd->dummy_netdev = kcalloc_node(1, netdev_size, GFP_KERNEL, dd->node);

and this kludgy code to find the rx data given a net_device:

static inline
struct hfi1_netdev_priv *hfi1_netdev_priv(struct net_device *dev)
{
        return (struct hfi1_netdev_priv *)&dev[1];
}

The solution is to embed the dummy net_device in a renamed hfi1_netdev_rx struct like other use cases.

The lifetime of that struct IS controlled properly and a kfree() of the renamed struct would include the embedded dummy.

Mike

[PATCH for-rc 2/2] IB/hfi1: Fix module use count flaw due to leftover module put calls

[Dennis Dalessandro]

When the try_module_get calls were removed from opening and closing of
the i2c debugfs file, the corresponding module_put calls were missed.
This results in an inaccurate module use count that requires a power
cycle to fix.

Fixes: 09fbca8e6240 ("IB/hfi1: No need to use try_module_get for debugfs")
Cc: <stable@vger.kernel.org>
Reviewed-by: Kaike Wan <kaike.wan@intel.com>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
---
 drivers/infiniband/hw/hfi1/debugfs.c |   19 ++-----------------
 1 file changed, 2 insertions(+), 17 deletions(-)

diff --git a/drivers/infiniband/hw/hfi1/debugfs.c b/drivers/infiniband/hw/hfi1/debugfs.c
index 4633a0c..2ced236 100644
--- a/drivers/infiniband/hw/hfi1/debugfs.c
+++ b/drivers/infiniband/hw/hfi1/debugfs.c
@@ -985,15 +985,10 @@ static ssize_t qsfp2_debugfs_read(struct file *file, char __user *buf,
 static int __i2c_debugfs_open(struct inode *in, struct file *fp, u32 target)
 {
 	struct hfi1_pportdata *ppd;
-	int ret;
 
 	ppd = private2ppd(fp);
 
-	ret = acquire_chip_resource(ppd->dd, i2c_target(target), 0);
-	if (ret) /* failed - release the module */
-		module_put(THIS_MODULE);
-
-	return ret;
+	return acquire_chip_resource(ppd->dd, i2c_target(target), 0);
 }
 
 static int i2c1_debugfs_open(struct inode *in, struct file *fp)
@@ -1013,7 +1008,6 @@ static int __i2c_debugfs_release(struct inode *in, struct file *fp, u32 target)
 	ppd = private2ppd(fp);
 
 	release_chip_resource(ppd->dd, i2c_target(target));
-	module_put(THIS_MODULE);
 
 	return 0;
 }
@@ -1031,18 +1025,10 @@ static int i2c2_debugfs_release(struct inode *in, struct file *fp)
 static int __qsfp_debugfs_open(struct inode *in, struct file *fp, u32 target)
 {
 	struct hfi1_pportdata *ppd;
-	int ret;
-
-	if (!try_module_get(THIS_MODULE))
-		return -ENODEV;
 
 	ppd = private2ppd(fp);
 
-	ret = acquire_chip_resource(ppd->dd, i2c_target(target), 0);
-	if (ret) /* failed - release the module */
-		module_put(THIS_MODULE);
-
-	return ret;
+	return acquire_chip_resource(ppd->dd, i2c_target(target), 0);
 }
 
 static int qsfp1_debugfs_open(struct inode *in, struct file *fp)
@@ -1062,7 +1048,6 @@ static int __qsfp_debugfs_release(struct inode *in, struct file *fp, u32 target)
 	ppd = private2ppd(fp);
 
 	release_chip_resource(ppd->dd, i2c_target(target));
-	module_put(THIS_MODULE);
 
 	return 0;
 }

Re: [PATCH for-rc 0/2] Fixes for module unload

[Jason Gunthorpe]

On Tue, Jun 23, 2020 at 04:32:18PM -0400, Dennis Dalessandro wrote:
> Two fixes for unload path. One is where the module use count got messed up in a
> previous fix, we missed hitting these paths. I missed hitting these paths. The
> other patch is going back to kfree for the dummy netdev. We are currently
> re-working how this dummy netdev stuff works and will send a series for-next
> soon.
> 
> Dennis Dalessandro (2):
>       IB/hfi1: Restore kfree in dummy_netdev cleanup
>       IB/hfi1: Fix module use count flaw due to leftover module put calls

I fixed the bogus Fixes line, please be more carefull

Applied to for-rc

Thanks,
Jason

Re: [PATCH for-rc 0/2] Fixes for module unload

[Dennis Dalessandro]

On 6/24/2020 3:01 PM, Jason Gunthorpe wrote:
> On Tue, Jun 23, 2020 at 04:32:18PM -0400, Dennis Dalessandro wrote:
>> Two fixes for unload path. One is where the module use count got messed up in a
>> previous fix, we missed hitting these paths. I missed hitting these paths. The
>> other patch is going back to kfree for the dummy netdev. We are currently
>> re-working how this dummy netdev stuff works and will send a series for-next
>> soon.
>>
>> Dennis Dalessandro (2):
>>        IB/hfi1: Restore kfree in dummy_netdev cleanup
>>        IB/hfi1: Fix module use count flaw due to leftover module put calls
> 
> I fixed the bogus Fixes line, please be more carefull
> 
> Applied to for-rc
>

Ah, I see the bug in my presubmit checks script. Will catch it next time.

-Denny