mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management
@ 2026-02-26  8:41 Koichiro Den
  2026-02-26  8:41 ` [PATCH v5 1/5] PCI: endpoint: pci-epf-vntb: Remove duplicate resource teardown Koichiro Den
                   ` (6 more replies)
  0 siblings, 7 replies; 14+ messages in thread
From: Koichiro Den @ 2026-02-26  8:41 UTC (permalink / raw)
  To: Manivannan Sadhasivam, Frank Li, Jon Mason, Dave Jiang,
	Allen Hubbe, kwilczynski, Kishon Vijay Abraham I, Bjorn Helgaas,
	Geert Uytterhoeven, Ye Bin, Lorenzo Pieralisi
  Cc: linux-pci, ntb, linux-kernel

The vNTB endpoint function (pci-epf-vntb) can be configured and
reconfigured through configfs (link/unlink functions, start/stop the
controller, update parameters). In practice, several pitfalls present:
duplicate EPC teardown that leads to oopses, a work item running after
resources were torn down, and inability to re-link/restart fundamentally
because ntb_dev was embedded and the vPCI bus teardown was incomplete.

This series addresses those issues and hardens resource management of
pci-epf-vntb:

- Remove duplicate EPC resource teardown in both pci-epf-vntb and
  pci-epf-ntb, avoiding crashes on .allow_link failures and during
  .drop_link.
- Stop the delayed cmd_handler work before clearing BARs/doorbells.
- Manage ntb_dev as a devm-managed allocation and implement .remove() in
  the vNTB PCI driver. Switch to pci_scan_root_bus().

With these changes, the controller can now be stopped, a function
unlinked, configfs settings updated, and the controller re-linked and
restarted without rebooting the endpoint, as long as the underlying
pci_epc_ops .stop() is non-destructive and .start() restores normal
operation.

Patches 1-3 carry Fixes tags and are candidates for stable.
Patch 4 is a preparatory one for Patch 5.
Patch 5 is a behavioral improvement that completes lifetime management for
relink/restart scenarios.

---
v4->v5 changes:
  - Rebased onto the latest pci/endpoint (2026-02-26).
  - Dropped [PATCH v4 1/7]; will be reposted separately via the NTB tree.
  - Dropped [PATCH v4 2/7], which has been applied in a different form.
  - Corrected the subject prefix of [PATCH v4 5/7]:
    s/NTB: epf: vntb:/PCI: endpoint: pci-epf-vntb:/.
  - Picked up a Reviewed-by tag to [PATCH v4 7/7].
  - Resolved a conflict in [PATCH v4 7/7] due to commit
    dc693d606644 ("PCI: endpoint: pci-epf-vntb: Add MSI doorbell support").
v3->v4 changes:
  - Added Reviewed-by tag for [PATCH v3 6/6].
  - Corrected patch split by moving the blank-line cleanup,
    based on the feedback from Frank.
  (No code changes overall.)
v2->v3 changes:
  - Added Reviewed-by tag for [PATCH v2 4/6].
  - Split [PATCH v2 6/6] into two, based on the feedback from Frank.
  (No code changes overall.)
v1->v2 changes:
  - Incorporated feedback from Frank.
  - Added Reviewed-by tags (except for patches #4 and #6).
  - Fixed a typo in patch #5 title.
  (No code changes overall.)

v4: https://lore.kernel.org/linux-pci/20251202072348.2752371-1-den@valinux.co.jp/
v3: https://lore.kernel.org/all/20251130151100.2591822-1-den@valinux.co.jp/
v2: https://lore.kernel.org/all/20251029080321.807943-1-den@valinux.co.jp/
v1: https://lore.kernel.org/all/20251023071757.901181-1-den@valinux.co.jp/


Koichiro Den (5):
  PCI: endpoint: pci-epf-vntb: Remove duplicate resource teardown
  PCI: endpoint: pci-epf-ntb: Remove duplicate resource teardown
  PCI: endpoint: pci-epf-vntb: Stop cmd_handler work in
    epf_ntb_epc_cleanup
  PCI: endpoint: pci-epf-vntb: Switch vpci_scan_bus() to use
    pci_scan_root_bus()
  PCI: endpoint: pci-epf-vntb: manage ntb_dev lifetime and fix vpci bus
    teardown

 drivers/pci/endpoint/functions/pci-epf-ntb.c  | 56 +-----------
 drivers/pci/endpoint/functions/pci-epf-vntb.c | 88 ++++++++++++-------
 2 files changed, 57 insertions(+), 87 deletions(-)

-- 
2.51.0


^ permalink raw reply	[flat|nested] 14+ messages in thread

* [PATCH v5 1/5] PCI: endpoint: pci-epf-vntb: Remove duplicate resource teardown
  2026-02-26  8:41 [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management Koichiro Den
@ 2026-02-26  8:41 ` Koichiro Den
  2026-02-26  8:41 ` [PATCH v5 2/5] PCI: endpoint: pci-epf-ntb: " Koichiro Den
                   ` (5 subsequent siblings)
  6 siblings, 0 replies; 14+ messages in thread
From: Koichiro Den @ 2026-02-26  8:41 UTC (permalink / raw)
  To: Manivannan Sadhasivam, Frank Li, Jon Mason, Dave Jiang,
	Allen Hubbe, kwilczynski, Kishon Vijay Abraham I, Bjorn Helgaas,
	Geert Uytterhoeven, Ye Bin, Lorenzo Pieralisi
  Cc: linux-pci, ntb, linux-kernel

epf_ntb_epc_destroy() duplicates the teardown that the caller is
supposed to perform later. This leads to an oops when .allow_link fails
or when .drop_link is performed. The following is an example oops of the
former case:

  Unable to handle kernel paging request at virtual address dead000000000108
  [...]
  [dead000000000108] address between user and kernel address ranges
  Internal error: Oops: 0000000096000044 [#1]  SMP
  [...]
  Call trace:
   pci_epc_remove_epf+0x78/0xe0 (P)
   pci_primary_epc_epf_link+0x88/0xa8
   configfs_symlink+0x1f4/0x5a0
   vfs_symlink+0x134/0x1d8
   do_symlinkat+0x88/0x138
   __arm64_sys_symlinkat+0x74/0xe0
  [...]

Remove the helper, and drop pci_epc_put(). EPC device refcounting is
tied to the configfs EPC group lifetime, and pci_epc_put() in the
.drop_link path is sufficient.

Cc: stable@vger.kernel.org
Fixes: e35f56bb0330 ("PCI: endpoint: Support NTB transfer between RC and EP")
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Koichiro Den <den@valinux.co.jp>
---
 drivers/pci/endpoint/functions/pci-epf-vntb.c | 19 +------------------
 1 file changed, 1 insertion(+), 18 deletions(-)

diff --git a/drivers/pci/endpoint/functions/pci-epf-vntb.c b/drivers/pci/endpoint/functions/pci-epf-vntb.c
index 148a3b160812..42c870ee3956 100644
--- a/drivers/pci/endpoint/functions/pci-epf-vntb.c
+++ b/drivers/pci/endpoint/functions/pci-epf-vntb.c
@@ -763,19 +763,6 @@ static void epf_ntb_mw_bar_clear(struct epf_ntb *ntb, int num_mws)
 	}
 }
 
-/**
- * epf_ntb_epc_destroy() - Cleanup NTB EPC interface
- * @ntb: NTB device that facilitates communication between HOST and VHOST
- *
- * Wrapper for epf_ntb_epc_destroy_interface() to cleanup all the NTB interfaces
- */
-static void epf_ntb_epc_destroy(struct epf_ntb *ntb)
-{
-	pci_epc_remove_epf(ntb->epf->epc, ntb->epf, 0);
-	pci_epc_put(ntb->epf->epc);
-}
-
-
 /**
  * epf_ntb_is_bar_used() - Check if a bar is used in the ntb configuration
  * @ntb: NTB device that facilitates communication between HOST and VHOST
@@ -1529,7 +1516,7 @@ static int epf_ntb_bind(struct pci_epf *epf)
 	ret = epf_ntb_init_epc_bar(ntb);
 	if (ret) {
 		dev_err(dev, "Failed to create NTB EPC\n");
-		goto err_bar_init;
+		return ret;
 	}
 
 	ret = epf_ntb_config_spad_bar_alloc(ntb);
@@ -1569,9 +1556,6 @@ static int epf_ntb_bind(struct pci_epf *epf)
 err_bar_alloc:
 	epf_ntb_config_spad_bar_free(ntb);
 
-err_bar_init:
-	epf_ntb_epc_destroy(ntb);
-
 	return ret;
 }
 
@@ -1587,7 +1571,6 @@ static void epf_ntb_unbind(struct pci_epf *epf)
 
 	epf_ntb_epc_cleanup(ntb);
 	epf_ntb_config_spad_bar_free(ntb);
-	epf_ntb_epc_destroy(ntb);
 
 	pci_unregister_driver(&vntb_pci_driver);
 }
-- 
2.51.0


^ permalink raw reply	[flat|nested] 14+ messages in thread

* [PATCH v5 2/5] PCI: endpoint: pci-epf-ntb: Remove duplicate resource teardown
  2026-02-26  8:41 [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management Koichiro Den
  2026-02-26  8:41 ` [PATCH v5 1/5] PCI: endpoint: pci-epf-vntb: Remove duplicate resource teardown Koichiro Den
@ 2026-02-26  8:41 ` Koichiro Den
  2026-02-26  8:41 ` [PATCH v5 3/5] PCI: endpoint: pci-epf-vntb: Stop cmd_handler work in epf_ntb_epc_cleanup Koichiro Den
                   ` (4 subsequent siblings)
  6 siblings, 0 replies; 14+ messages in thread
From: Koichiro Den @ 2026-02-26  8:41 UTC (permalink / raw)
  To: Manivannan Sadhasivam, Frank Li, Jon Mason, Dave Jiang,
	Allen Hubbe, kwilczynski, Kishon Vijay Abraham I, Bjorn Helgaas,
	Geert Uytterhoeven, Ye Bin, Lorenzo Pieralisi
  Cc: linux-pci, ntb, linux-kernel

epf_ntb_epc_destroy() duplicates the teardown that the caller is
supposed to do later. This leads to an oops when .allow_link fails or
when .drop_link is performed. Remove the helper.

Also drop pci_epc_put(). EPC device refcounting is tied to configfs EPC
group lifetime, and pci_epc_put() in the .drop_link path is sufficient.

Cc: stable@vger.kernel.org
Fixes: 8b821cf76150 ("PCI: endpoint: Add EP function driver to provide NTB functionality")
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Koichiro Den <den@valinux.co.jp>
---
 drivers/pci/endpoint/functions/pci-epf-ntb.c | 56 +-------------------
 1 file changed, 2 insertions(+), 54 deletions(-)

diff --git a/drivers/pci/endpoint/functions/pci-epf-ntb.c b/drivers/pci/endpoint/functions/pci-epf-ntb.c
index a3a588e522e7..2bdcc35b652c 100644
--- a/drivers/pci/endpoint/functions/pci-epf-ntb.c
+++ b/drivers/pci/endpoint/functions/pci-epf-ntb.c
@@ -1494,47 +1494,6 @@ static int epf_ntb_db_mw_bar_init(struct epf_ntb *ntb,
 	return ret;
 }
 
-/**
- * epf_ntb_epc_destroy_interface() - Cleanup NTB EPC interface
- * @ntb: NTB device that facilitates communication between HOST1 and HOST2
- * @type: PRIMARY interface or SECONDARY interface
- *
- * Unbind NTB function device from EPC and relinquish reference to pci_epc
- * for each of the interface.
- */
-static void epf_ntb_epc_destroy_interface(struct epf_ntb *ntb,
-					  enum pci_epc_interface_type type)
-{
-	struct epf_ntb_epc *ntb_epc;
-	struct pci_epc *epc;
-	struct pci_epf *epf;
-
-	if (type < 0)
-		return;
-
-	epf = ntb->epf;
-	ntb_epc = ntb->epc[type];
-	if (!ntb_epc)
-		return;
-	epc = ntb_epc->epc;
-	pci_epc_remove_epf(epc, epf, type);
-	pci_epc_put(epc);
-}
-
-/**
- * epf_ntb_epc_destroy() - Cleanup NTB EPC interface
- * @ntb: NTB device that facilitates communication between HOST1 and HOST2
- *
- * Wrapper for epf_ntb_epc_destroy_interface() to cleanup all the NTB interfaces
- */
-static void epf_ntb_epc_destroy(struct epf_ntb *ntb)
-{
-	enum pci_epc_interface_type type;
-
-	for (type = PRIMARY_INTERFACE; type <= SECONDARY_INTERFACE; type++)
-		epf_ntb_epc_destroy_interface(ntb, type);
-}
-
 /**
  * epf_ntb_epc_create_interface() - Create and initialize NTB EPC interface
  * @ntb: NTB device that facilitates communication between HOST1 and HOST2
@@ -1614,15 +1573,8 @@ static int epf_ntb_epc_create(struct epf_ntb *ntb)
 
 	ret = epf_ntb_epc_create_interface(ntb, epf->sec_epc,
 					   SECONDARY_INTERFACE);
-	if (ret) {
+	if (ret)
 		dev_err(dev, "SECONDARY intf: Fail to create NTB EPC\n");
-		goto err_epc_create;
-	}
-
-	return 0;
-
-err_epc_create:
-	epf_ntb_epc_destroy_interface(ntb, PRIMARY_INTERFACE);
 
 	return ret;
 }
@@ -1887,7 +1839,7 @@ static int epf_ntb_bind(struct pci_epf *epf)
 	ret = epf_ntb_init_epc_bar(ntb);
 	if (ret) {
 		dev_err(dev, "Failed to create NTB EPC\n");
-		goto err_bar_init;
+		return ret;
 	}
 
 	ret = epf_ntb_config_spad_bar_alloc_interface(ntb);
@@ -1909,9 +1861,6 @@ static int epf_ntb_bind(struct pci_epf *epf)
 err_bar_alloc:
 	epf_ntb_config_spad_bar_free(ntb);
 
-err_bar_init:
-	epf_ntb_epc_destroy(ntb);
-
 	return ret;
 }
 
@@ -1927,7 +1876,6 @@ static void epf_ntb_unbind(struct pci_epf *epf)
 
 	epf_ntb_epc_cleanup(ntb);
 	epf_ntb_config_spad_bar_free(ntb);
-	epf_ntb_epc_destroy(ntb);
 }
 
 #define EPF_NTB_R(_name)						\
-- 
2.51.0


^ permalink raw reply	[flat|nested] 14+ messages in thread

* [PATCH v5 3/5] PCI: endpoint: pci-epf-vntb: Stop cmd_handler work in epf_ntb_epc_cleanup
  2026-02-26  8:41 [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management Koichiro Den
  2026-02-26  8:41 ` [PATCH v5 1/5] PCI: endpoint: pci-epf-vntb: Remove duplicate resource teardown Koichiro Den
  2026-02-26  8:41 ` [PATCH v5 2/5] PCI: endpoint: pci-epf-ntb: " Koichiro Den
@ 2026-02-26  8:41 ` Koichiro Den
  2026-02-26  8:41 ` [PATCH v5 4/5] PCI: endpoint: pci-epf-vntb: Switch vpci_scan_bus() to use pci_scan_root_bus() Koichiro Den
                   ` (3 subsequent siblings)
  6 siblings, 0 replies; 14+ messages in thread
From: Koichiro Den @ 2026-02-26  8:41 UTC (permalink / raw)
  To: Manivannan Sadhasivam, Frank Li, Jon Mason, Dave Jiang,
	Allen Hubbe, kwilczynski, Kishon Vijay Abraham I, Bjorn Helgaas,
	Geert Uytterhoeven, Ye Bin, Lorenzo Pieralisi
  Cc: linux-pci, ntb, linux-kernel

Disable the delayed work before clearing BAR mappings and doorbells to
avoid running the handler after resources have been torn down.

  Unable to handle kernel paging request at virtual address ffff800083f46004
  [...]
  Internal error: Oops: 0000000096000007 [#1]  SMP
  [...]
  Call trace:
   epf_ntb_cmd_handler+0x54/0x200 [pci_epf_vntb] (P)
   process_one_work+0x154/0x3b0
   worker_thread+0x2c8/0x400
   kthread+0x148/0x210
   ret_from_fork+0x10/0x20

Cc: stable@vger.kernel.org
Fixes: e35f56bb0330 ("PCI: endpoint: Support NTB transfer between RC and EP")
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Koichiro Den <den@valinux.co.jp>
---
 drivers/pci/endpoint/functions/pci-epf-vntb.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/pci/endpoint/functions/pci-epf-vntb.c b/drivers/pci/endpoint/functions/pci-epf-vntb.c
index 42c870ee3956..805353528967 100644
--- a/drivers/pci/endpoint/functions/pci-epf-vntb.c
+++ b/drivers/pci/endpoint/functions/pci-epf-vntb.c
@@ -942,6 +942,7 @@ static int epf_ntb_epc_init(struct epf_ntb *ntb)
  */
 static void epf_ntb_epc_cleanup(struct epf_ntb *ntb)
 {
+	disable_delayed_work_sync(&ntb->cmd_handler);
 	epf_ntb_mw_bar_clear(ntb, ntb->num_mws);
 	epf_ntb_db_bar_clear(ntb);
 	epf_ntb_config_sspad_bar_clear(ntb);
-- 
2.51.0


^ permalink raw reply	[flat|nested] 14+ messages in thread

* [PATCH v5 4/5] PCI: endpoint: pci-epf-vntb: Switch vpci_scan_bus() to use pci_scan_root_bus()
  2026-02-26  8:41 [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management Koichiro Den
                   ` (2 preceding siblings ...)
  2026-02-26  8:41 ` [PATCH v5 3/5] PCI: endpoint: pci-epf-vntb: Stop cmd_handler work in epf_ntb_epc_cleanup Koichiro Den
@ 2026-02-26  8:41 ` Koichiro Den
  2026-03-02 16:09   ` Manivannan Sadhasivam
  2026-02-26  8:41 ` [PATCH v5 5/5] PCI: endpoint: pci-epf-vntb: manage ntb_dev lifetime and fix vpci bus teardown Koichiro Den
                   ` (2 subsequent siblings)
  6 siblings, 1 reply; 14+ messages in thread
From: Koichiro Den @ 2026-02-26  8:41 UTC (permalink / raw)
  To: Manivannan Sadhasivam, Frank Li, Jon Mason, Dave Jiang,
	Allen Hubbe, kwilczynski, Kishon Vijay Abraham I, Bjorn Helgaas,
	Geert Uytterhoeven, Ye Bin, Lorenzo Pieralisi
  Cc: linux-pci, ntb, linux-kernel

vpci_scan_bus() currently uses pci_scan_bus(), which creates a root bus
without a parent struct device. In a subsequent change we want to tear
down the virtual PCI root bus using pci_remove_root_bus(). For that to
work correctly, the root bus must be associated with a parent device,
similar to what the removed pci_scan_bus_parented() helper used to do.

Switch vpci_scan_bus() to use pci_scan_root_bus() and pass
&ndev->epf->epc->dev as the parent. Build the resource list in the same
way as pci_scan_bus(), so the behavior is unchanged except that the
virtual root bus now has a proper parent device. This avoids crashes in
the pci_epf_unbind() -> epf_ntb_unbind() -> pci_remove_root_bus() ->
pci_bus_release_domain_nr() path once we start removing the root bus in
a follow-up patch.

Reviewed-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Koichiro Den <den@valinux.co.jp>
---
Changes since v4:
  - Correct the subject prefix
    s/NTB: epf: vntb:/PCI: endpoint: pci-epf-vntb:/

 drivers/pci/endpoint/functions/pci-epf-vntb.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

diff --git a/drivers/pci/endpoint/functions/pci-epf-vntb.c b/drivers/pci/endpoint/functions/pci-epf-vntb.c
index 805353528967..f353e9a57194 100644
--- a/drivers/pci/endpoint/functions/pci-epf-vntb.c
+++ b/drivers/pci/endpoint/functions/pci-epf-vntb.c
@@ -1220,8 +1220,19 @@ static int vpci_scan_bus(void *sysdata)
 {
 	struct pci_bus *vpci_bus;
 	struct epf_ntb *ndev = sysdata;
-
-	vpci_bus = pci_scan_bus(ndev->vbus_number, &vpci_ops, sysdata);
+	LIST_HEAD(resources);
+	static struct resource busn_res = {
+		.start = 0,
+		.end = 255,
+		.flags = IORESOURCE_BUS,
+	};
+
+	pci_add_resource(&resources, &ioport_resource);
+	pci_add_resource(&resources, &iomem_resource);
+	pci_add_resource(&resources, &busn_res);
+
+	vpci_bus = pci_scan_root_bus(&ndev->epf->epc->dev, ndev->vbus_number,
+				     &vpci_ops, sysdata, &resources);
 	if (!vpci_bus) {
 		pr_err("create pci bus failed\n");
 		return -EINVAL;
-- 
2.51.0


^ permalink raw reply	[flat|nested] 14+ messages in thread

* [PATCH v5 5/5] PCI: endpoint: pci-epf-vntb: manage ntb_dev lifetime and fix vpci bus teardown
  2026-02-26  8:41 [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management Koichiro Den
                   ` (3 preceding siblings ...)
  2026-02-26  8:41 ` [PATCH v5 4/5] PCI: endpoint: pci-epf-vntb: Switch vpci_scan_bus() to use pci_scan_root_bus() Koichiro Den
@ 2026-02-26  8:41 ` Koichiro Den
  2026-03-03  5:22   ` Manivannan Sadhasivam
  2026-03-04  3:10 ` [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management Koichiro Den
  2026-03-04  6:41 ` (subset) " Manivannan Sadhasivam
  6 siblings, 1 reply; 14+ messages in thread
From: Koichiro Den @ 2026-02-26  8:41 UTC (permalink / raw)
  To: Manivannan Sadhasivam, Frank Li, Jon Mason, Dave Jiang,
	Allen Hubbe, kwilczynski, Kishon Vijay Abraham I, Bjorn Helgaas,
	Geert Uytterhoeven, Ye Bin, Lorenzo Pieralisi
  Cc: linux-pci, ntb, linux-kernel

Currently ntb_dev is embedded in epf_ntb, while configfs allows starting
or stopping controller and linking or unlinking functions as you want.
In fact, re-linking and re-starting is not possible with the embedded
design and leads to oopses.

Allocate ntb_dev with devm and add a .remove callback to the pci driver
that calls ntb_unregister_device(). This allows a fresh device to be
created on the next .bind call.

With these changes, the controller can now be stopped, a function
unlinked, configfs settings updated, and the controller re-linked and
restarted without rebooting the endpoint, as long as the underlying
pci_epc_ops .stop() operation is non-destructive, and .start() can
restore normal operations.

Reviewed-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Koichiro Den <den@valinux.co.jp>
---
Changes since v4:
  - Adjusted context due to commit:
    dc693d606644 ("PCI: endpoint: pci-epf-vntb: Add MSI doorbell support")

 drivers/pci/endpoint/functions/pci-epf-vntb.c | 53 ++++++++++++++-----
 1 file changed, 40 insertions(+), 13 deletions(-)

diff --git a/drivers/pci/endpoint/functions/pci-epf-vntb.c b/drivers/pci/endpoint/functions/pci-epf-vntb.c
index f353e9a57194..41a2f42e8a39 100644
--- a/drivers/pci/endpoint/functions/pci-epf-vntb.c
+++ b/drivers/pci/endpoint/functions/pci-epf-vntb.c
@@ -120,7 +120,7 @@ struct epf_ntb_ctrl {
 } __packed;
 
 struct epf_ntb {
-	struct ntb_dev ntb;
+	struct ntb_dev *ntb;
 	struct pci_epf *epf;
 	struct config_group group;
 
@@ -147,10 +147,16 @@ struct epf_ntb {
 	void __iomem *vpci_mw_addr[MAX_MW];
 
 	struct delayed_work cmd_handler;
+
+	struct pci_bus *vpci_bus;
 };
 
 #define to_epf_ntb(epf_group) container_of((epf_group), struct epf_ntb, group)
-#define ntb_ndev(__ntb) container_of(__ntb, struct epf_ntb, ntb)
+
+static inline struct epf_ntb *ntb_ndev(struct ntb_dev *ntb)
+{
+	return (struct epf_ntb *)ntb->pdev->sysdata;
+}
 
 static struct pci_epf_header epf_ntb_header = {
 	.vendorid	= PCI_ANY_ID,
@@ -176,7 +182,7 @@ static int epf_ntb_link_up(struct epf_ntb *ntb, bool link_up)
 	else
 		ntb->reg->link_status &= ~LINK_STATUS_UP;
 
-	ntb_link_event(&ntb->ntb);
+	ntb_link_event(ntb->ntb);
 	return 0;
 }
 
@@ -264,7 +270,7 @@ static void epf_ntb_cmd_handler(struct work_struct *work)
 	for (i = 1; i < ntb->db_count && !ntb->msi_doorbell; i++) {
 		if (ntb->epf_db[i]) {
 			atomic64_or(1 << (i - 1), &ntb->db);
-			ntb_db_event(&ntb->ntb, i);
+			ntb_db_event(ntb->ntb, i);
 			ntb->epf_db[i] = 0;
 		}
 	}
@@ -333,7 +339,7 @@ static irqreturn_t epf_ntb_doorbell_handler(int irq, void *data)
 	for (i = 1; i < ntb->db_count; i++)
 		if (irq == ntb->epf->db_msg[i].virq) {
 			atomic64_or(1 << (i - 1), &ntb->db);
-			ntb_db_event(&ntb->ntb, i);
+			ntb_db_event(ntb->ntb, i);
 		}
 
 	return IRQ_HANDLED;
@@ -1237,6 +1243,7 @@ static int vpci_scan_bus(void *sysdata)
 		pr_err("create pci bus failed\n");
 		return -EINVAL;
 	}
+	ndev->vpci_bus = vpci_bus;
 
 	pci_bus_add_devices(vpci_bus);
 
@@ -1281,7 +1288,7 @@ static int vntb_epf_mw_set_trans(struct ntb_dev *ndev, int pidx, int idx,
 	int ret;
 	struct device *dev;
 
-	dev = &ntb->ntb.dev;
+	dev = &ntb->ntb->dev;
 	barno = ntb->epf_ntb_bar[BAR_MW1 + idx];
 	epf_bar = &ntb->epf->bar[barno];
 	epf_bar->phys_addr = addr;
@@ -1381,7 +1388,7 @@ static int vntb_epf_peer_db_set(struct ntb_dev *ndev, u64 db_bits)
 	ret = pci_epc_raise_irq(ntb->epf->epc, func_no, vfunc_no,
 				PCI_IRQ_MSI, interrupt_num + 1);
 	if (ret)
-		dev_err(&ntb->ntb.dev, "Failed to raise IRQ\n");
+		dev_err(&ntb->ntb->dev, "Failed to raise IRQ\n");
 
 	return ret;
 }
@@ -1468,9 +1475,12 @@ static int pci_vntb_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	struct epf_ntb *ndev = (struct epf_ntb *)pdev->sysdata;
 	struct device *dev = &pdev->dev;
 
-	ndev->ntb.pdev = pdev;
-	ndev->ntb.topo = NTB_TOPO_NONE;
-	ndev->ntb.ops =  &vntb_epf_ops;
+	ndev->ntb = devm_kzalloc(dev, sizeof(*ndev->ntb), GFP_KERNEL);
+	if (!ndev->ntb)
+		return -ENOMEM;
+	ndev->ntb->pdev = pdev;
+	ndev->ntb->topo = NTB_TOPO_NONE;
+	ndev->ntb->ops = &vntb_epf_ops;
 
 	ret = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(32));
 	if (ret) {
@@ -1478,7 +1488,7 @@ static int pci_vntb_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 		return ret;
 	}
 
-	ret = ntb_register_device(&ndev->ntb);
+	ret = ntb_register_device(ndev->ntb);
 	if (ret) {
 		dev_err(dev, "Failed to register NTB device\n");
 		return ret;
@@ -1488,6 +1498,17 @@ static int pci_vntb_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	return 0;
 }
 
+static void pci_vntb_remove(struct pci_dev *pdev)
+{
+	struct epf_ntb *ndev = (struct epf_ntb *)pdev->sysdata;
+
+	if (!ndev || !ndev->ntb)
+		return;
+
+	ntb_unregister_device(ndev->ntb);
+	ndev->ntb = NULL;
+}
+
 static struct pci_device_id pci_vntb_table[] = {
 	{
 		PCI_DEVICE(0xffff, 0xffff),
@@ -1499,6 +1520,7 @@ static struct pci_driver vntb_pci_driver = {
 	.name           = "pci-vntb",
 	.id_table       = pci_vntb_table,
 	.probe          = pci_vntb_probe,
+	.remove         = pci_vntb_remove,
 };
 
 /* ============ PCIe EPF Driver Bind ====================*/
@@ -1581,10 +1603,15 @@ static void epf_ntb_unbind(struct pci_epf *epf)
 {
 	struct epf_ntb *ntb = epf_get_drvdata(epf);
 
+	pci_unregister_driver(&vntb_pci_driver);
+
+	pci_lock_rescan_remove();
+	pci_stop_root_bus(ntb->vpci_bus);
+	pci_remove_root_bus(ntb->vpci_bus);
+	pci_unlock_rescan_remove();
+
 	epf_ntb_epc_cleanup(ntb);
 	epf_ntb_config_spad_bar_free(ntb);
-
-	pci_unregister_driver(&vntb_pci_driver);
 }
 
 // EPF driver probe
-- 
2.51.0


^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: [PATCH v5 4/5] PCI: endpoint: pci-epf-vntb: Switch vpci_scan_bus() to use pci_scan_root_bus()
  2026-02-26  8:41 ` [PATCH v5 4/5] PCI: endpoint: pci-epf-vntb: Switch vpci_scan_bus() to use pci_scan_root_bus() Koichiro Den
@ 2026-03-02 16:09   ` Manivannan Sadhasivam
  2026-03-03  4:45     ` Koichiro Den
  0 siblings, 1 reply; 14+ messages in thread
From: Manivannan Sadhasivam @ 2026-03-02 16:09 UTC (permalink / raw)
  To: Koichiro Den
  Cc: Frank Li, Jon Mason, Dave Jiang, Allen Hubbe, kwilczynski,
	Kishon Vijay Abraham I, Bjorn Helgaas, Geert Uytterhoeven,
	Ye Bin, Lorenzo Pieralisi, linux-pci, ntb, linux-kernel

On Thu, Feb 26, 2026 at 05:41:41PM +0900, Koichiro Den wrote:
> vpci_scan_bus() currently uses pci_scan_bus(), which creates a root bus
> without a parent struct device. In a subsequent change we want to tear
> down the virtual PCI root bus using pci_remove_root_bus(). For that to
> work correctly, the root bus must be associated with a parent device,
> similar to what the removed pci_scan_bus_parented() helper used to do.
> 
> Switch vpci_scan_bus() to use pci_scan_root_bus() and pass
> &ndev->epf->epc->dev as the parent. Build the resource list in the same
> way as pci_scan_bus(), so the behavior is unchanged except that the
> virtual root bus now has a proper parent device. This avoids crashes in
> the pci_epf_unbind() -> epf_ntb_unbind() -> pci_remove_root_bus() ->
> pci_bus_release_domain_nr() path once we start removing the root bus in
> a follow-up patch.
> 
> Reviewed-by: Frank Li <Frank.Li@nxp.com>
> Signed-off-by: Koichiro Den <den@valinux.co.jp>
> ---
> Changes since v4:
>   - Correct the subject prefix
>     s/NTB: epf: vntb:/PCI: endpoint: pci-epf-vntb:/
> 
>  drivers/pci/endpoint/functions/pci-epf-vntb.c | 15 +++++++++++++--
>  1 file changed, 13 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/pci/endpoint/functions/pci-epf-vntb.c b/drivers/pci/endpoint/functions/pci-epf-vntb.c
> index 805353528967..f353e9a57194 100644
> --- a/drivers/pci/endpoint/functions/pci-epf-vntb.c
> +++ b/drivers/pci/endpoint/functions/pci-epf-vntb.c
> @@ -1220,8 +1220,19 @@ static int vpci_scan_bus(void *sysdata)
>  {
>  	struct pci_bus *vpci_bus;
>  	struct epf_ntb *ndev = sysdata;
> -
> -	vpci_bus = pci_scan_bus(ndev->vbus_number, &vpci_ops, sysdata);
> +	LIST_HEAD(resources);
> +	static struct resource busn_res = {
> +		.start = 0,
> +		.end = 255,
> +		.flags = IORESOURCE_BUS,
> +	};
> +
> +	pci_add_resource(&resources, &ioport_resource);
> +	pci_add_resource(&resources, &iomem_resource);
> +	pci_add_resource(&resources, &busn_res);
> +
> +	vpci_bus = pci_scan_root_bus(&ndev->epf->epc->dev, ndev->vbus_number,
> +				     &vpci_ops, sysdata, &resources);

Don't you need pci_lock_rescan_remove()/pci_unlock_rescan_remove()?

- Mani

-- 
மணிவண்ணன் சதாசிவம்

^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: [PATCH v5 4/5] PCI: endpoint: pci-epf-vntb: Switch vpci_scan_bus() to use pci_scan_root_bus()
  2026-03-02 16:09   ` Manivannan Sadhasivam
@ 2026-03-03  4:45     ` Koichiro Den
  0 siblings, 0 replies; 14+ messages in thread
From: Koichiro Den @ 2026-03-03  4:45 UTC (permalink / raw)
  To: Manivannan Sadhasivam
  Cc: Frank Li, Jon Mason, Dave Jiang, Allen Hubbe, kwilczynski,
	Kishon Vijay Abraham I, Bjorn Helgaas, Geert Uytterhoeven,
	Ye Bin, Lorenzo Pieralisi, linux-pci, ntb, linux-kernel

On Mon, Mar 02, 2026 at 09:39:26PM +0530, Manivannan Sadhasivam wrote:
> On Thu, Feb 26, 2026 at 05:41:41PM +0900, Koichiro Den wrote:
> > vpci_scan_bus() currently uses pci_scan_bus(), which creates a root bus
> > without a parent struct device. In a subsequent change we want to tear
> > down the virtual PCI root bus using pci_remove_root_bus(). For that to
> > work correctly, the root bus must be associated with a parent device,
> > similar to what the removed pci_scan_bus_parented() helper used to do.
> > 
> > Switch vpci_scan_bus() to use pci_scan_root_bus() and pass
> > &ndev->epf->epc->dev as the parent. Build the resource list in the same
> > way as pci_scan_bus(), so the behavior is unchanged except that the
> > virtual root bus now has a proper parent device. This avoids crashes in
> > the pci_epf_unbind() -> epf_ntb_unbind() -> pci_remove_root_bus() ->
> > pci_bus_release_domain_nr() path once we start removing the root bus in
> > a follow-up patch.
> > 
> > Reviewed-by: Frank Li <Frank.Li@nxp.com>
> > Signed-off-by: Koichiro Den <den@valinux.co.jp>
> > ---
> > Changes since v4:
> >   - Correct the subject prefix
> >     s/NTB: epf: vntb:/PCI: endpoint: pci-epf-vntb:/
> > 
> >  drivers/pci/endpoint/functions/pci-epf-vntb.c | 15 +++++++++++++--
> >  1 file changed, 13 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/pci/endpoint/functions/pci-epf-vntb.c b/drivers/pci/endpoint/functions/pci-epf-vntb.c
> > index 805353528967..f353e9a57194 100644
> > --- a/drivers/pci/endpoint/functions/pci-epf-vntb.c
> > +++ b/drivers/pci/endpoint/functions/pci-epf-vntb.c
> > @@ -1220,8 +1220,19 @@ static int vpci_scan_bus(void *sysdata)
> >  {
> >  	struct pci_bus *vpci_bus;
> >  	struct epf_ntb *ndev = sysdata;
> > -
> > -	vpci_bus = pci_scan_bus(ndev->vbus_number, &vpci_ops, sysdata);
> > +	LIST_HEAD(resources);
> > +	static struct resource busn_res = {
> > +		.start = 0,
> > +		.end = 255,
> > +		.flags = IORESOURCE_BUS,
> > +	};
> > +
> > +	pci_add_resource(&resources, &ioport_resource);
> > +	pci_add_resource(&resources, &iomem_resource);
> > +	pci_add_resource(&resources, &busn_res);
> > +
> > +	vpci_bus = pci_scan_root_bus(&ndev->epf->epc->dev, ndev->vbus_number,
> > +				     &vpci_ops, sysdata, &resources);
> 
> Don't you need pci_lock_rescan_remove()/pci_unlock_rescan_remove()?

Yes, we do need them. Thanks for pointing this out.

This locking was already missing before this series (when vpci_scan_bus() used
pci_scan_bus()), so I'll add a separate fix patch between Patch 3 and Patch 4,
likely with:
Fixes: e35f56bb0330 ("PCI: endpoint: Support NTB transfer between RC and EP").

---8<---
diff --git a/drivers/pci/endpoint/functions/pci-epf-vntb.c b/drivers/pci/endpoint/functions/pci-epf-vntb.c
index 805353528967..469b6e2b8d45 100644
--- a/drivers/pci/endpoint/functions/pci-epf-vntb.c
+++ b/drivers/pci/endpoint/functions/pci-epf-vntb.c
@@ -1221,14 +1221,19 @@ static int vpci_scan_bus(void *sysdata)
        struct pci_bus *vpci_bus;
        struct epf_ntb *ndev = sysdata;

+       pci_lock_rescan_remove();
+
        vpci_bus = pci_scan_bus(ndev->vbus_number, &vpci_ops, sysdata);
        if (!vpci_bus) {
                pr_err("create pci bus failed\n");
+               pci_unlock_rescan_remove();
                return -EINVAL;
        }

        pci_bus_add_devices(vpci_bus);

+       pci_unlock_rescan_remove();
+
        return 0;
 }

---8<---

Also, while respinning, I'll fix the error path in the pci_scan_root_bus()
version to free the temporary resource list when pci_scan_root_bus() fails,
since it doesn't free it on behalf of the caller. I noticed this while
revisiting the patch.

Best regards,
Koichiro

> 
> - Mani
> 
> -- 
> மணிவண்ணன் சதாசிவம்

^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: [PATCH v5 5/5] PCI: endpoint: pci-epf-vntb: manage ntb_dev lifetime and fix vpci bus teardown
  2026-02-26  8:41 ` [PATCH v5 5/5] PCI: endpoint: pci-epf-vntb: manage ntb_dev lifetime and fix vpci bus teardown Koichiro Den
@ 2026-03-03  5:22   ` Manivannan Sadhasivam
  2026-03-03 13:42     ` Koichiro Den
  0 siblings, 1 reply; 14+ messages in thread
From: Manivannan Sadhasivam @ 2026-03-03  5:22 UTC (permalink / raw)
  To: Koichiro Den
  Cc: Frank Li, Jon Mason, Dave Jiang, Allen Hubbe, kwilczynski,
	Kishon Vijay Abraham I, Bjorn Helgaas, Geert Uytterhoeven,
	Ye Bin, Lorenzo Pieralisi, linux-pci, ntb, linux-kernel

On Thu, Feb 26, 2026 at 05:41:42PM +0900, Koichiro Den wrote:
> Currently ntb_dev is embedded in epf_ntb, while configfs allows starting
> or stopping controller and linking or unlinking functions as you want.
> In fact, re-linking and re-starting is not possible with the embedded
> design and leads to oopses.
> 
> Allocate ntb_dev with devm and add a .remove callback to the pci driver
> that calls ntb_unregister_device(). This allows a fresh device to be
> created on the next .bind call.
> 
> With these changes, the controller can now be stopped, a function
> unlinked, configfs settings updated, and the controller re-linked and
> restarted without rebooting the endpoint, as long as the underlying
> pci_epc_ops .stop() operation is non-destructive, and .start() can
> restore normal operations.
> 
> Reviewed-by: Frank Li <Frank.Li@nxp.com>
> Signed-off-by: Koichiro Den <den@valinux.co.jp>
> ---
> Changes since v4:
>   - Adjusted context due to commit:
>     dc693d606644 ("PCI: endpoint: pci-epf-vntb: Add MSI doorbell support")
> 
>  drivers/pci/endpoint/functions/pci-epf-vntb.c | 53 ++++++++++++++-----
>  1 file changed, 40 insertions(+), 13 deletions(-)
> 
> diff --git a/drivers/pci/endpoint/functions/pci-epf-vntb.c b/drivers/pci/endpoint/functions/pci-epf-vntb.c
> index f353e9a57194..41a2f42e8a39 100644
> --- a/drivers/pci/endpoint/functions/pci-epf-vntb.c
> +++ b/drivers/pci/endpoint/functions/pci-epf-vntb.c
> @@ -120,7 +120,7 @@ struct epf_ntb_ctrl {
>  } __packed;
>  
>  struct epf_ntb {
> -	struct ntb_dev ntb;
> +	struct ntb_dev *ntb;
>  	struct pci_epf *epf;
>  	struct config_group group;
>  
> @@ -147,10 +147,16 @@ struct epf_ntb {
>  	void __iomem *vpci_mw_addr[MAX_MW];
>  
>  	struct delayed_work cmd_handler;
> +
> +	struct pci_bus *vpci_bus;
>  };
>  
>  #define to_epf_ntb(epf_group) container_of((epf_group), struct epf_ntb, group)
> -#define ntb_ndev(__ntb) container_of(__ntb, struct epf_ntb, ntb)
> +
> +static inline struct epf_ntb *ntb_ndev(struct ntb_dev *ntb)

Nit: No need of 'inline' keyword in a .c file.

- Mani

> +{
> +	return (struct epf_ntb *)ntb->pdev->sysdata;
> +}
>  
>  static struct pci_epf_header epf_ntb_header = {
>  	.vendorid	= PCI_ANY_ID,
> @@ -176,7 +182,7 @@ static int epf_ntb_link_up(struct epf_ntb *ntb, bool link_up)
>  	else
>  		ntb->reg->link_status &= ~LINK_STATUS_UP;
>  
> -	ntb_link_event(&ntb->ntb);
> +	ntb_link_event(ntb->ntb);
>  	return 0;
>  }
>  
> @@ -264,7 +270,7 @@ static void epf_ntb_cmd_handler(struct work_struct *work)
>  	for (i = 1; i < ntb->db_count && !ntb->msi_doorbell; i++) {
>  		if (ntb->epf_db[i]) {
>  			atomic64_or(1 << (i - 1), &ntb->db);
> -			ntb_db_event(&ntb->ntb, i);
> +			ntb_db_event(ntb->ntb, i);
>  			ntb->epf_db[i] = 0;
>  		}
>  	}
> @@ -333,7 +339,7 @@ static irqreturn_t epf_ntb_doorbell_handler(int irq, void *data)
>  	for (i = 1; i < ntb->db_count; i++)
>  		if (irq == ntb->epf->db_msg[i].virq) {
>  			atomic64_or(1 << (i - 1), &ntb->db);
> -			ntb_db_event(&ntb->ntb, i);
> +			ntb_db_event(ntb->ntb, i);
>  		}
>  
>  	return IRQ_HANDLED;
> @@ -1237,6 +1243,7 @@ static int vpci_scan_bus(void *sysdata)
>  		pr_err("create pci bus failed\n");
>  		return -EINVAL;
>  	}
> +	ndev->vpci_bus = vpci_bus;
>  
>  	pci_bus_add_devices(vpci_bus);
>  
> @@ -1281,7 +1288,7 @@ static int vntb_epf_mw_set_trans(struct ntb_dev *ndev, int pidx, int idx,
>  	int ret;
>  	struct device *dev;
>  
> -	dev = &ntb->ntb.dev;
> +	dev = &ntb->ntb->dev;
>  	barno = ntb->epf_ntb_bar[BAR_MW1 + idx];
>  	epf_bar = &ntb->epf->bar[barno];
>  	epf_bar->phys_addr = addr;
> @@ -1381,7 +1388,7 @@ static int vntb_epf_peer_db_set(struct ntb_dev *ndev, u64 db_bits)
>  	ret = pci_epc_raise_irq(ntb->epf->epc, func_no, vfunc_no,
>  				PCI_IRQ_MSI, interrupt_num + 1);
>  	if (ret)
> -		dev_err(&ntb->ntb.dev, "Failed to raise IRQ\n");
> +		dev_err(&ntb->ntb->dev, "Failed to raise IRQ\n");
>  
>  	return ret;
>  }
> @@ -1468,9 +1475,12 @@ static int pci_vntb_probe(struct pci_dev *pdev, const struct pci_device_id *id)
>  	struct epf_ntb *ndev = (struct epf_ntb *)pdev->sysdata;
>  	struct device *dev = &pdev->dev;
>  
> -	ndev->ntb.pdev = pdev;
> -	ndev->ntb.topo = NTB_TOPO_NONE;
> -	ndev->ntb.ops =  &vntb_epf_ops;
> +	ndev->ntb = devm_kzalloc(dev, sizeof(*ndev->ntb), GFP_KERNEL);
> +	if (!ndev->ntb)
> +		return -ENOMEM;
> +	ndev->ntb->pdev = pdev;
> +	ndev->ntb->topo = NTB_TOPO_NONE;
> +	ndev->ntb->ops = &vntb_epf_ops;
>  
>  	ret = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(32));
>  	if (ret) {
> @@ -1478,7 +1488,7 @@ static int pci_vntb_probe(struct pci_dev *pdev, const struct pci_device_id *id)
>  		return ret;
>  	}
>  
> -	ret = ntb_register_device(&ndev->ntb);
> +	ret = ntb_register_device(ndev->ntb);
>  	if (ret) {
>  		dev_err(dev, "Failed to register NTB device\n");
>  		return ret;
> @@ -1488,6 +1498,17 @@ static int pci_vntb_probe(struct pci_dev *pdev, const struct pci_device_id *id)
>  	return 0;
>  }
>  
> +static void pci_vntb_remove(struct pci_dev *pdev)
> +{
> +	struct epf_ntb *ndev = (struct epf_ntb *)pdev->sysdata;
> +
> +	if (!ndev || !ndev->ntb)
> +		return;
> +
> +	ntb_unregister_device(ndev->ntb);
> +	ndev->ntb = NULL;
> +}
> +
>  static struct pci_device_id pci_vntb_table[] = {
>  	{
>  		PCI_DEVICE(0xffff, 0xffff),
> @@ -1499,6 +1520,7 @@ static struct pci_driver vntb_pci_driver = {
>  	.name           = "pci-vntb",
>  	.id_table       = pci_vntb_table,
>  	.probe          = pci_vntb_probe,
> +	.remove         = pci_vntb_remove,
>  };
>  
>  /* ============ PCIe EPF Driver Bind ====================*/
> @@ -1581,10 +1603,15 @@ static void epf_ntb_unbind(struct pci_epf *epf)
>  {
>  	struct epf_ntb *ntb = epf_get_drvdata(epf);
>  
> +	pci_unregister_driver(&vntb_pci_driver);
> +
> +	pci_lock_rescan_remove();
> +	pci_stop_root_bus(ntb->vpci_bus);
> +	pci_remove_root_bus(ntb->vpci_bus);
> +	pci_unlock_rescan_remove();
> +
>  	epf_ntb_epc_cleanup(ntb);
>  	epf_ntb_config_spad_bar_free(ntb);
> -
> -	pci_unregister_driver(&vntb_pci_driver);
>  }
>  
>  // EPF driver probe
> -- 
> 2.51.0
> 

-- 
மணிவண்ணன் சதாசிவம்

^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: [PATCH v5 5/5] PCI: endpoint: pci-epf-vntb: manage ntb_dev lifetime and fix vpci bus teardown
  2026-03-03  5:22   ` Manivannan Sadhasivam
@ 2026-03-03 13:42     ` Koichiro Den
  0 siblings, 0 replies; 14+ messages in thread
From: Koichiro Den @ 2026-03-03 13:42 UTC (permalink / raw)
  To: Manivannan Sadhasivam
  Cc: Frank Li, Jon Mason, Dave Jiang, Allen Hubbe, kwilczynski,
	Kishon Vijay Abraham I, Bjorn Helgaas, Geert Uytterhoeven,
	Ye Bin, Lorenzo Pieralisi, linux-pci, ntb, linux-kernel

On Tue, Mar 03, 2026 at 10:52:05AM +0530, Manivannan Sadhasivam wrote:
> On Thu, Feb 26, 2026 at 05:41:42PM +0900, Koichiro Den wrote:
> > Currently ntb_dev is embedded in epf_ntb, while configfs allows starting
> > or stopping controller and linking or unlinking functions as you want.
> > In fact, re-linking and re-starting is not possible with the embedded
> > design and leads to oopses.
> > 
> > Allocate ntb_dev with devm and add a .remove callback to the pci driver
> > that calls ntb_unregister_device(). This allows a fresh device to be
> > created on the next .bind call.
> > 
> > With these changes, the controller can now be stopped, a function
> > unlinked, configfs settings updated, and the controller re-linked and
> > restarted without rebooting the endpoint, as long as the underlying
> > pci_epc_ops .stop() operation is non-destructive, and .start() can
> > restore normal operations.
> > 
> > Reviewed-by: Frank Li <Frank.Li@nxp.com>
> > Signed-off-by: Koichiro Den <den@valinux.co.jp>
> > ---
> > Changes since v4:
> >   - Adjusted context due to commit:
> >     dc693d606644 ("PCI: endpoint: pci-epf-vntb: Add MSI doorbell support")
> > 
> >  drivers/pci/endpoint/functions/pci-epf-vntb.c | 53 ++++++++++++++-----
> >  1 file changed, 40 insertions(+), 13 deletions(-)
> > 
> > diff --git a/drivers/pci/endpoint/functions/pci-epf-vntb.c b/drivers/pci/endpoint/functions/pci-epf-vntb.c
> > index f353e9a57194..41a2f42e8a39 100644
> > --- a/drivers/pci/endpoint/functions/pci-epf-vntb.c
> > +++ b/drivers/pci/endpoint/functions/pci-epf-vntb.c
> > @@ -120,7 +120,7 @@ struct epf_ntb_ctrl {
> >  } __packed;
> >  
> >  struct epf_ntb {
> > -	struct ntb_dev ntb;
> > +	struct ntb_dev *ntb;
> >  	struct pci_epf *epf;
> >  	struct config_group group;
> >  
> > @@ -147,10 +147,16 @@ struct epf_ntb {
> >  	void __iomem *vpci_mw_addr[MAX_MW];
> >  
> >  	struct delayed_work cmd_handler;
> > +
> > +	struct pci_bus *vpci_bus;
> >  };
> >  
> >  #define to_epf_ntb(epf_group) container_of((epf_group), struct epf_ntb, group)
> > -#define ntb_ndev(__ntb) container_of(__ntb, struct epf_ntb, ntb)
> > +
> > +static inline struct epf_ntb *ntb_ndev(struct ntb_dev *ntb)
> 
> Nit: No need of 'inline' keyword in a .c file.

Thanks for catching this. I'll drop the 'inline' in v6.


While revisiting Patch 5/5 carefully, I also noticed a lifetime issue around
cmd_handler (and doorbell IRQ handlers).

1. Race window between pci_unregister_driver() and work/IRQ teardown

   With this patch, the following window exists:

   -> epf_ntb_unbind()
      -> pci_unregister_driver()
         ...
         -> pci_vntb_remove()
            -> ntb_unregister_device()
            -> ndev->ntb = NULL ----------------------------- (A)
      -> epf_ntb_epc_cleanup()
         -> disable_delayed_work_sync(&ntb->cmd_handler) ---- (B)
         -> epf_ntb_db_bar_clear() (free_irq, etc.) --------- (C)

   If cmd_handler runs between (A) and (B) (or if epf_ntb_doorbell_handler()
   runs between (A) and (C)), they may call ntb_{db,link}_event(ndev->ntb, ...)
   with ndev->ntb == NULL, leading to a NULL deref.  Also, since the doorbell
   IRQs are requested during bind, the IRQ handler can run even before
   pci_vntb_probe() ever sets up ndev->ntb.

   In current mainline, cmd_handler is initially scheduled in
   epf_ntb_epc_init(), and the doorbell IRQs are also requested during bind,
   i.e. both are set up before pci_register_driver() and ndev->ntb becomes
   valid.

   I think the cleanest solution is to tie cmd_handler and the "enabled" state
   of the doorbell IRQs to the lifetime of struct ntb_dev:

   - Move the initial scheduling (queue_work(..., &ndev->cmd_handler.work)) to
     pci_vntb_probe(), after ntb_register_device() succeeds, and stop it in
     pci_vntb_remove() using disable_delayed_work_sync() before
     ntb_unregister_device().

   - Keep request_irq()/free_irq() in the existing bind/unbind paths, but gate
     delivery in vPCI probe/remove: keep the doorbell IRQs disabled until
     pci_vntb_probe() succeeds (enable_irq() after ndev->ntb becomes valid), and
     disable_irq() them in pci_vntb_remove() before clearing ndev->ntb /
     ntb_unregister_device().

   That way, the unsafe windows I wrote above disappear, and the lifetime of
   struct ntb_dev and its associated resources are managed symmetrically across
   bind/unbind and probe/remove paths.

2. Stale ndev->ntb on probe failure

   pci_vntb_probe() assigns ndev->ntb before ntb_register_device(). If probe
   fails after devm_kzalloc(), ndev->ntb may still hold a stale pointer. I'll
   fix this by avoiding publishing ndev->ntb until ntb_register_device()
   succeeds (or clear it on the error path).

I'll include these fixes into v6.

Best regards,
Koichiro

> 
> - Mani
> 
> > +{
> > +	return (struct epf_ntb *)ntb->pdev->sysdata;
> > +}
> >  
> >  static struct pci_epf_header epf_ntb_header = {
> >  	.vendorid	= PCI_ANY_ID,
> > @@ -176,7 +182,7 @@ static int epf_ntb_link_up(struct epf_ntb *ntb, bool link_up)
> >  	else
> >  		ntb->reg->link_status &= ~LINK_STATUS_UP;
> >  
> > -	ntb_link_event(&ntb->ntb);
> > +	ntb_link_event(ntb->ntb);
> >  	return 0;
> >  }
> >  
> > @@ -264,7 +270,7 @@ static void epf_ntb_cmd_handler(struct work_struct *work)
> >  	for (i = 1; i < ntb->db_count && !ntb->msi_doorbell; i++) {
> >  		if (ntb->epf_db[i]) {
> >  			atomic64_or(1 << (i - 1), &ntb->db);
> > -			ntb_db_event(&ntb->ntb, i);
> > +			ntb_db_event(ntb->ntb, i);
> >  			ntb->epf_db[i] = 0;
> >  		}
> >  	}
> > @@ -333,7 +339,7 @@ static irqreturn_t epf_ntb_doorbell_handler(int irq, void *data)
> >  	for (i = 1; i < ntb->db_count; i++)
> >  		if (irq == ntb->epf->db_msg[i].virq) {
> >  			atomic64_or(1 << (i - 1), &ntb->db);
> > -			ntb_db_event(&ntb->ntb, i);
> > +			ntb_db_event(ntb->ntb, i);
> >  		}
> >  
> >  	return IRQ_HANDLED;
> > @@ -1237,6 +1243,7 @@ static int vpci_scan_bus(void *sysdata)
> >  		pr_err("create pci bus failed\n");
> >  		return -EINVAL;
> >  	}
> > +	ndev->vpci_bus = vpci_bus;
> >  
> >  	pci_bus_add_devices(vpci_bus);
> >  
> > @@ -1281,7 +1288,7 @@ static int vntb_epf_mw_set_trans(struct ntb_dev *ndev, int pidx, int idx,
> >  	int ret;
> >  	struct device *dev;
> >  
> > -	dev = &ntb->ntb.dev;
> > +	dev = &ntb->ntb->dev;
> >  	barno = ntb->epf_ntb_bar[BAR_MW1 + idx];
> >  	epf_bar = &ntb->epf->bar[barno];
> >  	epf_bar->phys_addr = addr;
> > @@ -1381,7 +1388,7 @@ static int vntb_epf_peer_db_set(struct ntb_dev *ndev, u64 db_bits)
> >  	ret = pci_epc_raise_irq(ntb->epf->epc, func_no, vfunc_no,
> >  				PCI_IRQ_MSI, interrupt_num + 1);
> >  	if (ret)
> > -		dev_err(&ntb->ntb.dev, "Failed to raise IRQ\n");
> > +		dev_err(&ntb->ntb->dev, "Failed to raise IRQ\n");
> >  
> >  	return ret;
> >  }
> > @@ -1468,9 +1475,12 @@ static int pci_vntb_probe(struct pci_dev *pdev, const struct pci_device_id *id)
> >  	struct epf_ntb *ndev = (struct epf_ntb *)pdev->sysdata;
> >  	struct device *dev = &pdev->dev;
> >  
> > -	ndev->ntb.pdev = pdev;
> > -	ndev->ntb.topo = NTB_TOPO_NONE;
> > -	ndev->ntb.ops =  &vntb_epf_ops;
> > +	ndev->ntb = devm_kzalloc(dev, sizeof(*ndev->ntb), GFP_KERNEL);
> > +	if (!ndev->ntb)
> > +		return -ENOMEM;
> > +	ndev->ntb->pdev = pdev;
> > +	ndev->ntb->topo = NTB_TOPO_NONE;
> > +	ndev->ntb->ops = &vntb_epf_ops;
> >  
> >  	ret = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(32));
> >  	if (ret) {
> > @@ -1478,7 +1488,7 @@ static int pci_vntb_probe(struct pci_dev *pdev, const struct pci_device_id *id)
> >  		return ret;
> >  	}
> >  
> > -	ret = ntb_register_device(&ndev->ntb);
> > +	ret = ntb_register_device(ndev->ntb);
> >  	if (ret) {
> >  		dev_err(dev, "Failed to register NTB device\n");
> >  		return ret;
> > @@ -1488,6 +1498,17 @@ static int pci_vntb_probe(struct pci_dev *pdev, const struct pci_device_id *id)
> >  	return 0;
> >  }
> >  
> > +static void pci_vntb_remove(struct pci_dev *pdev)
> > +{
> > +	struct epf_ntb *ndev = (struct epf_ntb *)pdev->sysdata;
> > +
> > +	if (!ndev || !ndev->ntb)
> > +		return;
> > +
> > +	ntb_unregister_device(ndev->ntb);
> > +	ndev->ntb = NULL;
> > +}
> > +
> >  static struct pci_device_id pci_vntb_table[] = {
> >  	{
> >  		PCI_DEVICE(0xffff, 0xffff),
> > @@ -1499,6 +1520,7 @@ static struct pci_driver vntb_pci_driver = {
> >  	.name           = "pci-vntb",
> >  	.id_table       = pci_vntb_table,
> >  	.probe          = pci_vntb_probe,
> > +	.remove         = pci_vntb_remove,
> >  };
> >  
> >  /* ============ PCIe EPF Driver Bind ====================*/
> > @@ -1581,10 +1603,15 @@ static void epf_ntb_unbind(struct pci_epf *epf)
> >  {
> >  	struct epf_ntb *ntb = epf_get_drvdata(epf);
> >  
> > +	pci_unregister_driver(&vntb_pci_driver);
> > +
> > +	pci_lock_rescan_remove();
> > +	pci_stop_root_bus(ntb->vpci_bus);
> > +	pci_remove_root_bus(ntb->vpci_bus);
> > +	pci_unlock_rescan_remove();
> > +
> >  	epf_ntb_epc_cleanup(ntb);
> >  	epf_ntb_config_spad_bar_free(ntb);
> > -
> > -	pci_unregister_driver(&vntb_pci_driver);
> >  }
> >  
> >  // EPF driver probe
> > -- 
> > 2.51.0
> > 
> 
> -- 
> மணிவண்ணன் சதாசிவம்

^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management
  2026-02-26  8:41 [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management Koichiro Den
                   ` (4 preceding siblings ...)
  2026-02-26  8:41 ` [PATCH v5 5/5] PCI: endpoint: pci-epf-vntb: manage ntb_dev lifetime and fix vpci bus teardown Koichiro Den
@ 2026-03-04  3:10 ` Koichiro Den
  2026-03-04  6:33   ` Manivannan Sadhasivam
  2026-03-04  6:41 ` (subset) " Manivannan Sadhasivam
  6 siblings, 1 reply; 14+ messages in thread
From: Koichiro Den @ 2026-03-04  3:10 UTC (permalink / raw)
  To: Manivannan Sadhasivam, Frank Li, Jon Mason, Dave Jiang,
	Allen Hubbe, kwilczynski, Kishon Vijay Abraham I, Bjorn Helgaas,
	Geert Uytterhoeven, Ye Bin, Lorenzo Pieralisi
  Cc: linux-pci, ntb, linux-kernel

On Thu, Feb 26, 2026 at 05:41:37PM +0900, Koichiro Den wrote:
> The vNTB endpoint function (pci-epf-vntb) can be configured and
> reconfigured through configfs (link/unlink functions, start/stop the
> controller, update parameters). In practice, several pitfalls present:
> duplicate EPC teardown that leads to oopses, a work item running after
> resources were torn down, and inability to re-link/restart fundamentally
> because ntb_dev was embedded and the vPCI bus teardown was incomplete.
> 
> This series addresses those issues and hardens resource management of
> pci-epf-vntb:
> 
> - Remove duplicate EPC resource teardown in both pci-epf-vntb and
>   pci-epf-ntb, avoiding crashes on .allow_link failures and during
>   .drop_link.
> - Stop the delayed cmd_handler work before clearing BARs/doorbells.
> - Manage ntb_dev as a devm-managed allocation and implement .remove() in
>   the vNTB PCI driver. Switch to pci_scan_root_bus().
> 
> With these changes, the controller can now be stopped, a function
> unlinked, configfs settings updated, and the controller re-linked and
> restarted without rebooting the endpoint, as long as the underlying
> pci_epc_ops .stop() is non-destructive and .start() restores normal
> operation.
> 
> Patches 1-3 carry Fixes tags and are candidates for stable.
> Patch 4 is a preparatory one for Patch 5.
> Patch 5 is a behavioral improvement that completes lifetime management for
> relink/restart scenarios.

While I'm updating Patch 4 and 5 to address feedback from Mani, as well as the
concern I mentioned at [1], I noticed that if [2] gets merged before this
series, another issue may arrise. With [2], the DB IRQ may become a shared IRQ,
in which case the unbind/remove race would require additional care.


Mani, if it's ok, could you take Patch 1-3?

- If so, I'll spin the rest (Patch 4-5) into a separate patch series starting
  from v6, with some additional commits.

  It turns out that Patch 4-5 are a bigger change than I initially thought. Even
  though Patch 1-3 were originally written as preparatory fixes, they can be
  applied independently at any time.

  The code in Patches 1-3 has also been unchanged since v1 (submitted last
  October).

[1] https://lore.kernel.org/linux-pci/mipdls67csyyrugf4rjx3qqtbxes4sjjtluy3psecnadcgcs7k@rn42d3m6ggsf/
[2] [PATCH v10 0/7] PCI: endpoint: pci-ep-msi: Add embedded doorbell fallback
    https://lore.kernel.org/linux-pci/20260302071427.534158-1-den@valinux.co.jp/


Best regards,
Koichiro

> 
> ---
> v4->v5 changes:
>   - Rebased onto the latest pci/endpoint (2026-02-26).
>   - Dropped [PATCH v4 1/7]; will be reposted separately via the NTB tree.
>   - Dropped [PATCH v4 2/7], which has been applied in a different form.
>   - Corrected the subject prefix of [PATCH v4 5/7]:
>     s/NTB: epf: vntb:/PCI: endpoint: pci-epf-vntb:/.
>   - Picked up a Reviewed-by tag to [PATCH v4 7/7].
>   - Resolved a conflict in [PATCH v4 7/7] due to commit
>     dc693d606644 ("PCI: endpoint: pci-epf-vntb: Add MSI doorbell support").
> v3->v4 changes:
>   - Added Reviewed-by tag for [PATCH v3 6/6].
>   - Corrected patch split by moving the blank-line cleanup,
>     based on the feedback from Frank.
>   (No code changes overall.)
> v2->v3 changes:
>   - Added Reviewed-by tag for [PATCH v2 4/6].
>   - Split [PATCH v2 6/6] into two, based on the feedback from Frank.
>   (No code changes overall.)
> v1->v2 changes:
>   - Incorporated feedback from Frank.
>   - Added Reviewed-by tags (except for patches #4 and #6).
>   - Fixed a typo in patch #5 title.
>   (No code changes overall.)
> 
> v4: https://lore.kernel.org/linux-pci/20251202072348.2752371-1-den@valinux.co.jp/
> v3: https://lore.kernel.org/all/20251130151100.2591822-1-den@valinux.co.jp/
> v2: https://lore.kernel.org/all/20251029080321.807943-1-den@valinux.co.jp/
> v1: https://lore.kernel.org/all/20251023071757.901181-1-den@valinux.co.jp/
> 
> 
> Koichiro Den (5):
>   PCI: endpoint: pci-epf-vntb: Remove duplicate resource teardown
>   PCI: endpoint: pci-epf-ntb: Remove duplicate resource teardown
>   PCI: endpoint: pci-epf-vntb: Stop cmd_handler work in
>     epf_ntb_epc_cleanup
>   PCI: endpoint: pci-epf-vntb: Switch vpci_scan_bus() to use
>     pci_scan_root_bus()
>   PCI: endpoint: pci-epf-vntb: manage ntb_dev lifetime and fix vpci bus
>     teardown
> 
>  drivers/pci/endpoint/functions/pci-epf-ntb.c  | 56 +-----------
>  drivers/pci/endpoint/functions/pci-epf-vntb.c | 88 ++++++++++++-------
>  2 files changed, 57 insertions(+), 87 deletions(-)
> 
> -- 
> 2.51.0
> 
> 

^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management
  2026-03-04  3:10 ` [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management Koichiro Den
@ 2026-03-04  6:33   ` Manivannan Sadhasivam
  0 siblings, 0 replies; 14+ messages in thread
From: Manivannan Sadhasivam @ 2026-03-04  6:33 UTC (permalink / raw)
  To: Koichiro Den
  Cc: Frank Li, Jon Mason, Dave Jiang, Allen Hubbe, kwilczynski,
	Kishon Vijay Abraham I, Bjorn Helgaas, Geert Uytterhoeven,
	Ye Bin, Lorenzo Pieralisi, linux-pci, ntb, linux-kernel

On Wed, Mar 04, 2026 at 12:10:23PM +0900, Koichiro Den wrote:
> On Thu, Feb 26, 2026 at 05:41:37PM +0900, Koichiro Den wrote:
> > The vNTB endpoint function (pci-epf-vntb) can be configured and
> > reconfigured through configfs (link/unlink functions, start/stop the
> > controller, update parameters). In practice, several pitfalls present:
> > duplicate EPC teardown that leads to oopses, a work item running after
> > resources were torn down, and inability to re-link/restart fundamentally
> > because ntb_dev was embedded and the vPCI bus teardown was incomplete.
> > 
> > This series addresses those issues and hardens resource management of
> > pci-epf-vntb:
> > 
> > - Remove duplicate EPC resource teardown in both pci-epf-vntb and
> >   pci-epf-ntb, avoiding crashes on .allow_link failures and during
> >   .drop_link.
> > - Stop the delayed cmd_handler work before clearing BARs/doorbells.
> > - Manage ntb_dev as a devm-managed allocation and implement .remove() in
> >   the vNTB PCI driver. Switch to pci_scan_root_bus().
> > 
> > With these changes, the controller can now be stopped, a function
> > unlinked, configfs settings updated, and the controller re-linked and
> > restarted without rebooting the endpoint, as long as the underlying
> > pci_epc_ops .stop() is non-destructive and .start() restores normal
> > operation.
> > 
> > Patches 1-3 carry Fixes tags and are candidates for stable.
> > Patch 4 is a preparatory one for Patch 5.
> > Patch 5 is a behavioral improvement that completes lifetime management for
> > relink/restart scenarios.
> 
> While I'm updating Patch 4 and 5 to address feedback from Mani, as well as the
> concern I mentioned at [1], I noticed that if [2] gets merged before this
> series, another issue may arrise. With [2], the DB IRQ may become a shared IRQ,
> in which case the unbind/remove race would require additional care.
> 
> 
> Mani, if it's ok, could you take Patch 1-3?
> 
> - If so, I'll spin the rest (Patch 4-5) into a separate patch series starting
>   from v6, with some additional commits.
> 

Sounds OK to me.

- Mani

>   It turns out that Patch 4-5 are a bigger change than I initially thought. Even
>   though Patch 1-3 were originally written as preparatory fixes, they can be
>   applied independently at any time.
> 
>   The code in Patches 1-3 has also been unchanged since v1 (submitted last
>   October).
> 
> [1] https://lore.kernel.org/linux-pci/mipdls67csyyrugf4rjx3qqtbxes4sjjtluy3psecnadcgcs7k@rn42d3m6ggsf/
> [2] [PATCH v10 0/7] PCI: endpoint: pci-ep-msi: Add embedded doorbell fallback
>     https://lore.kernel.org/linux-pci/20260302071427.534158-1-den@valinux.co.jp/
> 
> 
> Best regards,
> Koichiro
> 
> > 
> > ---
> > v4->v5 changes:
> >   - Rebased onto the latest pci/endpoint (2026-02-26).
> >   - Dropped [PATCH v4 1/7]; will be reposted separately via the NTB tree.
> >   - Dropped [PATCH v4 2/7], which has been applied in a different form.
> >   - Corrected the subject prefix of [PATCH v4 5/7]:
> >     s/NTB: epf: vntb:/PCI: endpoint: pci-epf-vntb:/.
> >   - Picked up a Reviewed-by tag to [PATCH v4 7/7].
> >   - Resolved a conflict in [PATCH v4 7/7] due to commit
> >     dc693d606644 ("PCI: endpoint: pci-epf-vntb: Add MSI doorbell support").
> > v3->v4 changes:
> >   - Added Reviewed-by tag for [PATCH v3 6/6].
> >   - Corrected patch split by moving the blank-line cleanup,
> >     based on the feedback from Frank.
> >   (No code changes overall.)
> > v2->v3 changes:
> >   - Added Reviewed-by tag for [PATCH v2 4/6].
> >   - Split [PATCH v2 6/6] into two, based on the feedback from Frank.
> >   (No code changes overall.)
> > v1->v2 changes:
> >   - Incorporated feedback from Frank.
> >   - Added Reviewed-by tags (except for patches #4 and #6).
> >   - Fixed a typo in patch #5 title.
> >   (No code changes overall.)
> > 
> > v4: https://lore.kernel.org/linux-pci/20251202072348.2752371-1-den@valinux.co.jp/
> > v3: https://lore.kernel.org/all/20251130151100.2591822-1-den@valinux.co.jp/
> > v2: https://lore.kernel.org/all/20251029080321.807943-1-den@valinux.co.jp/
> > v1: https://lore.kernel.org/all/20251023071757.901181-1-den@valinux.co.jp/
> > 
> > 
> > Koichiro Den (5):
> >   PCI: endpoint: pci-epf-vntb: Remove duplicate resource teardown
> >   PCI: endpoint: pci-epf-ntb: Remove duplicate resource teardown
> >   PCI: endpoint: pci-epf-vntb: Stop cmd_handler work in
> >     epf_ntb_epc_cleanup
> >   PCI: endpoint: pci-epf-vntb: Switch vpci_scan_bus() to use
> >     pci_scan_root_bus()
> >   PCI: endpoint: pci-epf-vntb: manage ntb_dev lifetime and fix vpci bus
> >     teardown
> > 
> >  drivers/pci/endpoint/functions/pci-epf-ntb.c  | 56 +-----------
> >  drivers/pci/endpoint/functions/pci-epf-vntb.c | 88 ++++++++++++-------
> >  2 files changed, 57 insertions(+), 87 deletions(-)
> > 
> > -- 
> > 2.51.0
> > 
> > 

-- 
மணிவண்ணன் சதாசிவம்

^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: (subset) [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management
  2026-02-26  8:41 [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management Koichiro Den
                   ` (5 preceding siblings ...)
  2026-03-04  3:10 ` [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management Koichiro Den
@ 2026-03-04  6:41 ` Manivannan Sadhasivam
  2026-03-04  8:43   ` Koichiro Den
  6 siblings, 1 reply; 14+ messages in thread
From: Manivannan Sadhasivam @ 2026-03-04  6:41 UTC (permalink / raw)
  To: Frank Li, Jon Mason, Dave Jiang, Allen Hubbe, kwilczynski,
	Kishon Vijay Abraham I, Bjorn Helgaas, Geert Uytterhoeven,
	Ye Bin, Lorenzo Pieralisi, Koichiro Den
  Cc: linux-pci, ntb, linux-kernel


On Thu, 26 Feb 2026 17:41:37 +0900, Koichiro Den wrote:
> The vNTB endpoint function (pci-epf-vntb) can be configured and
> reconfigured through configfs (link/unlink functions, start/stop the
> controller, update parameters). In practice, several pitfalls present:
> duplicate EPC teardown that leads to oopses, a work item running after
> resources were torn down, and inability to re-link/restart fundamentally
> because ntb_dev was embedded and the vPCI bus teardown was incomplete.
> 
> [...]

Applied, thanks!

[1/5] PCI: endpoint: pci-epf-vntb: Remove duplicate resource teardown
      commit: 0da63230d3ec1ec5fcc443a2314233e95bfece54
[2/5] PCI: endpoint: pci-epf-ntb: Remove duplicate resource teardown
      commit: 3446beddba450c8d6f9aca2f028712ac527fead3
[3/5] PCI: endpoint: pci-epf-vntb: Stop cmd_handler work in epf_ntb_epc_cleanup
      commit: d799984233a50abd2667a7d17a9a710a3f10ebe2

Best regards,
-- 
Manivannan Sadhasivam <mani@kernel.org>


^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: (subset) [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management
  2026-03-04  6:41 ` (subset) " Manivannan Sadhasivam
@ 2026-03-04  8:43   ` Koichiro Den
  0 siblings, 0 replies; 14+ messages in thread
From: Koichiro Den @ 2026-03-04  8:43 UTC (permalink / raw)
  To: Manivannan Sadhasivam
  Cc: Frank Li, Jon Mason, Dave Jiang, Allen Hubbe, kwilczynski,
	Kishon Vijay Abraham I, Bjorn Helgaas, Geert Uytterhoeven,
	Ye Bin, Lorenzo Pieralisi, linux-pci, ntb, linux-kernel

On Wed, Mar 04, 2026 at 12:11:11PM +0530, Manivannan Sadhasivam wrote:
> 
> On Thu, 26 Feb 2026 17:41:37 +0900, Koichiro Den wrote:
> > The vNTB endpoint function (pci-epf-vntb) can be configured and
> > reconfigured through configfs (link/unlink functions, start/stop the
> > controller, update parameters). In practice, several pitfalls present:
> > duplicate EPC teardown that leads to oopses, a work item running after
> > resources were torn down, and inability to re-link/restart fundamentally
> > because ntb_dev was embedded and the vPCI bus teardown was incomplete.
> > 
> > [...]
> 
> Applied, thanks!
> 
> [1/5] PCI: endpoint: pci-epf-vntb: Remove duplicate resource teardown
>       commit: 0da63230d3ec1ec5fcc443a2314233e95bfece54
> [2/5] PCI: endpoint: pci-epf-ntb: Remove duplicate resource teardown
>       commit: 3446beddba450c8d6f9aca2f028712ac527fead3
> [3/5] PCI: endpoint: pci-epf-vntb: Stop cmd_handler work in epf_ntb_epc_cleanup
>       commit: d799984233a50abd2667a7d17a9a710a3f10ebe2

Thanks for taking the subset, Mani.

I'll prepare the remaining patches as a separate series after taking another
careful look.

Best regards,
Koichiro

> 
> Best regards,
> -- 
> Manivannan Sadhasivam <mani@kernel.org>
> 

^ permalink raw reply	[flat|nested] 14+ messages in thread

end of thread, other threads:[~2026-03-04  8:43 UTC | newest]

Thread overview: 14+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-02-26  8:41 [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management Koichiro Den
2026-02-26  8:41 ` [PATCH v5 1/5] PCI: endpoint: pci-epf-vntb: Remove duplicate resource teardown Koichiro Den
2026-02-26  8:41 ` [PATCH v5 2/5] PCI: endpoint: pci-epf-ntb: " Koichiro Den
2026-02-26  8:41 ` [PATCH v5 3/5] PCI: endpoint: pci-epf-vntb: Stop cmd_handler work in epf_ntb_epc_cleanup Koichiro Den
2026-02-26  8:41 ` [PATCH v5 4/5] PCI: endpoint: pci-epf-vntb: Switch vpci_scan_bus() to use pci_scan_root_bus() Koichiro Den
2026-03-02 16:09   ` Manivannan Sadhasivam
2026-03-03  4:45     ` Koichiro Den
2026-02-26  8:41 ` [PATCH v5 5/5] PCI: endpoint: pci-epf-vntb: manage ntb_dev lifetime and fix vpci bus teardown Koichiro Den
2026-03-03  5:22   ` Manivannan Sadhasivam
2026-03-03 13:42     ` Koichiro Den
2026-03-04  3:10 ` [PATCH v5 0/5] PCI: endpoint: pci-epf-*ntb: Harden vNTB resource management Koichiro Den
2026-03-04  6:33   ` Manivannan Sadhasivam
2026-03-04  6:41 ` (subset) " Manivannan Sadhasivam
2026-03-04  8:43   ` Koichiro Den

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®