* Re: [PATCH net-next] net: mana: Force full-page RX buffers for 4K page size on specific systems.
From: Dipayaan Roy @ 2026-03-06 13:12 UTC (permalink / raw)
To: Paolo Abeni
Cc: kys, haiyangz, wei.liu, decui, andrew+netdev, davem, edumazet,
kuba, leon, longli, kotaranov, horms, shradhagupta, ssengar,
ernis, shirazsaleem, linux-hyperv, netdev, linux-kernel,
linux-rdma, dipayanroy
In-Reply-To: <81b7e296-0cfe-4c24-ac97-7f6c712aa0e9@redhat.com>
On Tue, Mar 03, 2026 at 11:56:29AM +0100, Paolo Abeni wrote:
> On 2/27/26 11:15 AM, Dipayaan Roy wrote:
> > On certain systems configured with 4K PAGE_SIZE, utilizing page_pool
> > fragments for RX buffers results in a significant throughput regression.
> > Profiling reveals that this regression correlates with high overhead in the
> > fragment allocation and reference counting paths on these specific
> > platforms, rendering the multi-buffer-per-page strategy counterproductive.
> >
> > To mitigate this, bypass the page_pool fragment path and force a single RX
> > packet per page allocation when all the following conditions are met:
> > 1. The system is configured with a 4K PAGE_SIZE.
> > 2. A processor-specific quirk is detected via SMBIOS Type 4 data.
> >
> > This approach restores expected line-rate performance by ensuring
> > predictable RX refill behavior on affected hardware.
> >
> > There is no behavioral change for systems using larger page sizes
> > (16K/64K), or platforms where this processor-specific quirk do not
> > apply.
> >
> > Signed-off-by: Dipayaan Roy <dipayanroy@linux.microsoft.com>
> > ---
> > .../net/ethernet/microsoft/mana/gdma_main.c | 120 ++++++++++++++++++
> > drivers/net/ethernet/microsoft/mana/mana_en.c | 23 +++-
> > include/net/mana/gdma.h | 10 ++
> > 3 files changed, 151 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/net/ethernet/microsoft/mana/gdma_main.c b/drivers/net/ethernet/microsoft/mana/gdma_main.c
> > index 0055c231acf6..26bbe736a770 100644
> > --- a/drivers/net/ethernet/microsoft/mana/gdma_main.c
> > +++ b/drivers/net/ethernet/microsoft/mana/gdma_main.c
> > @@ -9,6 +9,7 @@
> > #include <linux/msi.h>
> > #include <linux/irqdomain.h>
> > #include <linux/export.h>
> > +#include <linux/dmi.h>
> >
> > #include <net/mana/mana.h>
> > #include <net/mana/hw_channel.h>
> > @@ -1955,6 +1956,115 @@ static bool mana_is_pf(unsigned short dev_id)
> > return dev_id == MANA_PF_DEVICE_ID;
> > }
> >
> > +/*
> > + * Table for Processor Version strings found from SMBIOS Type 4 information,
> > + * for processors that needs to force single RX buffer per page quirk for
> > + * meeting line rate performance with ARM64 + 4K pages.
> > + * Note: These strings are exactly matched with version fetched from SMBIOS.
> > + */
> > +static const char * const mana_single_rxbuf_per_page_quirk_tbl[] = {
> > + "Cobalt 200",
> > +};
> > +
> > +static const char *smbios_get_string(const struct dmi_header *hdr, u8 idx)
> > +{
> > + const u8 *start, *end;
> > + u8 i;
> > +
> > + /* Indexing starts from 1. */
> > + if (!idx)
> > + return NULL;
> > +
> > + start = (const u8 *)hdr + hdr->length;
> > + end = start + SMBIOS_STR_AREA_MAX;
> > +
> > + for (i = 1; i < idx; i++) {
> > + while (start < end && *start)
> > + start++;
> > + if (start < end)
> > + start++;
> > + if (start + 1 < end && start[0] == 0 && start[1] == 0)
> > + return NULL;
> > + }
> > +
> > + if (start >= end || *start == 0)
> > + return NULL;
> > +
> > + return (const char *)start;
>
> If I read correctly, the above sort of duplicate dmi_decode_table().
>
Yes, its not exported.
> I think you are better of:
> - use the mana_get_proc_ver_from_smbios() decoder to store the
> SMBIOS_TYPE4_PROC_VERSION_OFFSET index into gd
> - do a 2nd walk with a different decoder to fetch the string at the
> specified index.
Sure, will implement the 2nd walk for fetching string in v2.
>
> /P
Thank you Paolo, for the comments, and apologies in my delay in response as this week I am on-call.
I will send out v2 with the changes suggested.
Regards
^ permalink raw reply
* Re: [PATCH net-next] net: mana: Force full-page RX buffers for 4K page size on specific systems.
From: Dipayaan Roy @ 2026-03-06 13:25 UTC (permalink / raw)
To: Paolo Abeni
Cc: kys, haiyangz, wei.liu, decui, andrew+netdev, davem, edumazet,
kuba, leon, longli, kotaranov, horms, shradhagupta, ssengar,
ernis, shirazsaleem, linux-hyperv, netdev, linux-kernel,
linux-rdma, dipayanroy
In-Reply-To: <03ac38dc-69c5-4641-98ea-5679465c0b7f@redhat.com>
On Tue, Mar 03, 2026 at 12:56:35PM +0100, Paolo Abeni wrote:
> On 2/27/26 11:15 AM, Dipayaan Roy wrote:
> > diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
> > index 91c418097284..a53a8921050b 100644
> > --- a/drivers/net/ethernet/microsoft/mana/mana_en.c
> > +++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
> > @@ -748,6 +748,26 @@ static void *mana_get_rxbuf_pre(struct mana_rxq *rxq, dma_addr_t *da)
> > return va;
> > }
> >
> > +static inline bool
> > +mana_use_single_rxbuf_per_page(struct mana_port_context *apc, u32 mtu)
> > +{
>
> I almost forgot: please avoid the 'inline' keyword in .c files. This is
> function used only once, should be inlined by the compiler anyway.
>
Ack, will remove it in v2.
> > + struct gdma_context *gc = apc->ac->gdma_dev->gdma_context;
> > +
> > + /* On some systems with 4K PAGE_SIZE, page_pool RX fragments can
> > + * trigger a throughput regression. Hence forces one RX buffer per page
> > + * to avoid the fragment allocation/refcounting overhead in the RX
> > + * refill path for those processors only.
> > + */
> > + if (gc->force_full_page_rx_buffer)
> > + return true;
>
> Side note: since you could keep the above flag up2date according to the
> current mtu and xdp configuration and just test it in the data path.
>
If not an issue, would like to keep it this way for better readability.
> /P
>
Regrads
^ permalink raw reply
* [PATCH net-next, v2] net: mana: Force full-page RX buffers for 4K page size on specific systems.
From: Dipayaan Roy @ 2026-03-06 13:33 UTC (permalink / raw)
To: kys, haiyangz, wei.liu, decui, andrew+netdev, davem, edumazet,
kuba, pabeni, leon, longli, kotaranov, horms, shradhagupta,
ssengar, ernis, shirazsaleem, linux-hyperv, netdev, linux-kernel,
linux-rdma, dipayanroy
On certain systems configured with 4K PAGE_SIZE, utilizing page_pool
fragments for RX buffers results in a significant throughput regression.
Profiling reveals that this regression correlates with high overhead in the
fragment allocation and reference counting paths on these specific
platforms, rendering the multi-buffer-per-page strategy counterproductive.
To mitigate this, bypass the page_pool fragment path and force a single RX
packet per page allocation when all the following conditions are met:
1. The system is configured with a 4K PAGE_SIZE.
2. A processor-specific quirk is detected via SMBIOS Type 4 data.
This approach restores expected line-rate performance by ensuring
predictable RX refill behavior on affected hardware.
There is no behavioral change for systems using larger page sizes
(16K/64K), or platforms where this processor-specific quirk do not
apply.
Signed-off-by: Dipayaan Roy <dipayanroy@linux.microsoft.com>
---
Changes in v2:
- separate reading string index and the string, remove inline.
---
---
.../net/ethernet/microsoft/mana/gdma_main.c | 133 ++++++++++++++++++
drivers/net/ethernet/microsoft/mana/mana_en.c | 23 ++-
include/net/mana/gdma.h | 9 ++
3 files changed, 163 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/microsoft/mana/gdma_main.c b/drivers/net/ethernet/microsoft/mana/gdma_main.c
index aef8612b73cb..05fecc00a90c 100644
--- a/drivers/net/ethernet/microsoft/mana/gdma_main.c
+++ b/drivers/net/ethernet/microsoft/mana/gdma_main.c
@@ -9,6 +9,7 @@
#include <linux/msi.h>
#include <linux/irqdomain.h>
#include <linux/export.h>
+#include <linux/dmi.h>
#include <net/mana/mana.h>
#include <net/mana/hw_channel.h>
@@ -1959,6 +1960,128 @@ static bool mana_is_pf(unsigned short dev_id)
return dev_id == MANA_PF_DEVICE_ID;
}
+/*
+ * Table for Processor Version strings found from SMBIOS Type 4 information,
+ * for processors that needs to force single RX buffer per page quirk for
+ * meeting line rate performance with ARM64 + 4K pages.
+ * Note: These strings are exactly matched with version fetched from SMBIOS.
+ */
+static const char * const mana_single_rxbuf_per_page_quirk_tbl[] = {
+ "Cobalt 200",
+};
+
+/* On some systems with 4K PAGE_SIZE, page_pool RX fragments can
+ * trigger a throughput regression. Hence identify those processors
+ * from the extracted SMBIOS table and apply the quirk to forces one
+ * RX buffer per page to avoid the fragment allocation/refcounting
+ * overhead in the RX refill path for those processors only.
+ */
+static bool mana_needs_single_rxbuf_per_page(struct gdma_context *gc)
+{
+ int i = 0;
+ const char *ver = gc->processor_version;
+
+ if (!ver)
+ return false;
+
+ if (PAGE_SIZE != SZ_4K)
+ return false;
+
+ while (i < ARRAY_SIZE(mana_single_rxbuf_per_page_quirk_tbl)) {
+ if (!strcmp(ver, mana_single_rxbuf_per_page_quirk_tbl[i]))
+ return true;
+ i++;
+ }
+
+ return false;
+}
+
+static void mana_get_proc_ver_strno(const struct dmi_header *hdr, void *data)
+{
+ struct gdma_context *gc = data;
+ const u8 *d = (const u8 *)hdr;
+
+ /* We are only looking for Type 4: Processor Information */
+ if (hdr->type != SMBIOS_TYPE_4_PROCESSOR_INFO)
+ return;
+
+ /* Ensure the record is long enough to contain the Processor Version
+ * field
+ */
+ if (hdr->length <= SMBIOS_TYPE4_PROC_VERSION_OFFSET)
+ return;
+
+ /* The 'Processor Version' string is located at index pointed by
+ * SMBIOS_TYPE4_PROC_VERSION_OFFSET. Make a copy of the index.
+ * There could be multiple Type 4 tables so read and store the
+ * processor version index found the first time.
+ */
+ if (gc->proc_ver_strno)
+ return;
+
+ gc->proc_ver_strno = d[SMBIOS_TYPE4_PROC_VERSION_OFFSET];
+}
+
+static const char *mana_dmi_string_nosave(const struct dmi_header *hdr, u8 s)
+{
+ const char *bp = (const char *)hdr + hdr->length;
+
+ if (!s)
+ return NULL;
+
+ /* String numbers start at 1 */
+ while (--s > 0 && *bp)
+ bp += strlen(bp) + 1;
+
+ if (!*bp)
+ return NULL;
+
+ return bp;
+}
+
+static void mana_fetch_proc_ver_string(const struct dmi_header *hdr,
+ void *data)
+{
+ struct gdma_context *gc = data;
+ const char *ver;
+
+ /* We are only looking for Type 4: Processor Information */
+ if (hdr->type != SMBIOS_TYPE_4_PROCESSOR_INFO)
+ return;
+
+ /* Extract proc version found the first time only */
+ if (!gc->proc_ver_strno || gc->processor_version)
+ return;
+
+ ver = mana_dmi_string_nosave(hdr, gc->proc_ver_strno);
+ if (ver)
+ gc->processor_version = kstrdup(ver, GFP_KERNEL);
+}
+
+/* Check and initialize all processor optimizations/quirks here */
+static bool mana_init_processor_optimization(struct gdma_context *gc)
+{
+ bool opt_initialized = false;
+
+ gc->proc_ver_strno = 0;
+ gc->processor_version = NULL;
+
+ dmi_walk(mana_get_proc_ver_strno, gc);
+ if (!gc->proc_ver_strno)
+ return false;
+
+ dmi_walk(mana_fetch_proc_ver_string, gc);
+ if (!gc->processor_version)
+ return false;
+
+ if (mana_needs_single_rxbuf_per_page(gc)) {
+ gc->force_full_page_rx_buffer = true;
+ opt_initialized = true;
+ }
+
+ return opt_initialized;
+}
+
static int mana_gd_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
{
struct gdma_context *gc;
@@ -2013,6 +2136,11 @@ static int mana_gd_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
gc->mana_pci_debugfs = debugfs_create_dir(pci_slot_name(pdev->slot),
mana_debugfs_root);
+ if (mana_init_processor_optimization(gc))
+ dev_info(&pdev->dev,
+ "Processor specific optimization initialized on: %s\n",
+ gc->processor_version);
+
err = mana_gd_setup(pdev);
if (err)
goto unmap_bar;
@@ -2055,6 +2183,8 @@ static int mana_gd_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
pci_iounmap(pdev, bar0_va);
free_gc:
pci_set_drvdata(pdev, NULL);
+ kfree(gc->processor_version);
+ gc->processor_version = NULL;
vfree(gc);
release_region:
pci_release_regions(pdev);
@@ -2110,6 +2240,9 @@ static void mana_gd_remove(struct pci_dev *pdev)
pci_iounmap(pdev, gc->bar0_va);
+ kfree(gc->processor_version);
+ gc->processor_version = NULL;
+
vfree(gc);
pci_release_regions(pdev);
diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
index a868c28c8280..38f94f7619ad 100644
--- a/drivers/net/ethernet/microsoft/mana/mana_en.c
+++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
@@ -744,6 +744,26 @@ static void *mana_get_rxbuf_pre(struct mana_rxq *rxq, dma_addr_t *da)
return va;
}
+static bool
+mana_use_single_rxbuf_per_page(struct mana_port_context *apc, u32 mtu)
+{
+ struct gdma_context *gc = apc->ac->gdma_dev->gdma_context;
+
+ /* On some systems with 4K PAGE_SIZE, page_pool RX fragments can
+ * trigger a throughput regression. Hence forces one RX buffer per page
+ * to avoid the fragment allocation/refcounting overhead in the RX
+ * refill path for those processors only.
+ */
+ if (gc->force_full_page_rx_buffer)
+ return true;
+
+ /* For xdp and jumbo frames make sure only one packet fits per page. */
+ if (mtu + MANA_RXBUF_PAD > PAGE_SIZE / 2 || mana_xdp_get(apc))
+ return true;
+
+ return false;
+}
+
/* Get RX buffer's data size, alloc size, XDP headroom based on MTU */
static void mana_get_rxbuf_cfg(struct mana_port_context *apc,
int mtu, u32 *datasize, u32 *alloc_size,
@@ -754,8 +774,7 @@ static void mana_get_rxbuf_cfg(struct mana_port_context *apc,
/* Calculate datasize first (consistent across all cases) */
*datasize = mtu + ETH_HLEN;
- /* For xdp and jumbo frames make sure only one packet fits per page */
- if (mtu + MANA_RXBUF_PAD > PAGE_SIZE / 2 || mana_xdp_get(apc)) {
+ if (mana_use_single_rxbuf_per_page(apc, mtu)) {
if (mana_xdp_get(apc)) {
*headroom = XDP_PACKET_HEADROOM;
*alloc_size = PAGE_SIZE;
diff --git a/include/net/mana/gdma.h b/include/net/mana/gdma.h
index ec17004b10c0..be56b347f3f6 100644
--- a/include/net/mana/gdma.h
+++ b/include/net/mana/gdma.h
@@ -9,6 +9,12 @@
#include "shm_channel.h"
+/* SMBIOS Type 4: Processor Information table */
+#define SMBIOS_TYPE_4_PROCESSOR_INFO 4
+
+/* Byte offset containing the Processor Version string number.*/
+#define SMBIOS_TYPE4_PROC_VERSION_OFFSET 0x10
+
#define GDMA_STATUS_MORE_ENTRIES 0x00000105
#define GDMA_STATUS_CMD_UNSUPPORTED 0xffffffff
@@ -444,6 +450,9 @@ struct gdma_context {
struct workqueue_struct *service_wq;
unsigned long flags;
+ u8 *processor_version;
+ u8 proc_ver_strno;
+ bool force_full_page_rx_buffer;
};
static inline bool mana_gd_is_mana(struct gdma_dev *gd)
--
2.34.1
^ permalink raw reply related
* RE: [PATCH v2] hv_utils: Allow implicit ICTIMESYNCFLAG_SYNC
From: David Balazic @ 2026-03-06 14:25 UTC (permalink / raw)
To: pmartincic@linux.microsoft.com, K. Y. Srinivasan, Haiyang Zhang,
Wei Liu, Dexuan Cui, linux-hyperv@vger.kernel.org,
linux-kernel@vger.kernel.org
In-Reply-To: <20231127213524.52783-1-pmartincic@linux.microsoft.com>
On Mon, Nov 27, 2023 at 01:35:24PM -0800, pmartincic@linux.microsoft.com wrote:
> From: Peter Martincic <pmartincic@microsoft.com>
>
> Hyper-V hosts can omit the _SYNC flag to due a bug on resume from modern
> suspend. In such a case, the guest may fail to update its time-of-day to
> account for the period when it was suspended, and could proceed with a
> significantly wrong time-of-day. In such a case when the guest is significantly
> behind, fix it by treating a _SAMPLE the same as if _SYNC was received so that
> the guest time-of-day is updated.
>
> This is hidden behind param hv_utils.timesync_implicit.
>
> Signed-off-by: Peter Martincic <pmartincic@microsoft.com>
Hi!
As Peters mail does not seem to exists any more, I'll ask here (LKML and LHV), if all right:
Is there any news about this bug on Hyper-V hosts?
If this is actually a Hyper-V (host) bug, was it ever addressed and fixed?
I encounter this bug when using Windows 11 Enterprise (25H2) host and Oracle Linux 8 and 9 as guests.
They use an older kernel, so I can't use this parameter to fix it.
Windows guests do not seem to be affected by it, in my experience.
Feel free to reply off list, as it is not really on topic.
Regards,
David Balažic
The information in this email is confidential and may be legally privileged. It is intended solely for the addressee. Any opinions expressed are mine and do not necessarily represent the opinions of the Company. Emails are susceptible to interference. If you are not the intended recipient, any disclosure, copying, distribution or any action taken or omitted to be taken in reliance on it, is strictly prohibited and may be unlawful. If you have received this message in error, do not open any attachments but please notify the Endava Service Desk on (+44 (0)870 423 0187), and delete this message from your system. The sender accepts no responsibility for information, errors or omissions in this email, or for its use or misuse, or for any act committed or omitted in connection with this communication. If in doubt, please verify the authenticity of the contents with the sender. Please rely on your own virus checkers as no responsibility is taken by the sender for any damage rising out of any bug or virus infection.
Endava plc is a company registered in England under company number 5722669 whose registered office is at 125 Old Broad Street, London, EC2N 1AR, United Kingdom. Endava plc is the Endava group holding company and does not provide any services to clients. Each of Endava plc and its subsidiaries is a separate legal entity and has no liability for another such entity's acts or omissions.
^ permalink raw reply
* Re: [v2,net-next,1/6] net: mana: Create separate EQs for each vPort
From: Simon Horman @ 2026-03-06 16:08 UTC (permalink / raw)
To: longli
Cc: Simon Horman, davem, edumazet, haiyangz, linux-rdma, linux-hyperv,
kuba, wei.liu, shradhagupta, pabeni, kotaranov, schakrabarti,
erick.archer, linux-kernel, kys, decui, netdev
In-Reply-To: <20260304000017.333312-2-longli@microsoft.com>
This is an AI-generated review of your patch. The human sending this
email has considered the AI review valid, or at least plausible.
For local reproduction steps see https://netdev-ai.bots.linux.dev/ai-local.html
---
net: mana: Create separate EQs for each vPort
To prepare for assigning vPorts to dedicated MSI-X vectors, this commit
removes EQ sharing among vPorts and creates dedicated EQs for each vPort.
The EQ definition is moved from struct mana_context to struct
mana_port_context and related support functions are updated.
> diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h
> --- a/include/net/mana/mana.h
> +++ b/include/net/mana/mana.h
[ ... ]
> @@ -499,6 +497,9 @@ struct mana_port_context {
>
> u8 mac_addr[ETH_ALEN];
>
> + struct mana_eq *eqs;
> + struct dentry *mana_eqs_debugfs;
> +
> enum TRI_STATE rss_state;
>
> mana_handle_t default_rxobj;
[ ... ]
> diff --git a/drivers/infiniband/hw/mana/qp.c b/drivers/infiniband/hw/mana/qp.c
> --- a/drivers/infiniband/hw/mana/qp.c
> +++ b/drivers/infiniband/hw/mana/qp.c
> @@ -188,7 +188,7 @@ struct mana_obj_spec cq_spec = {};
> cq_spec.gdma_region = cq->queue.gdma_region;
> cq_spec.queue_size = cq->cqe * COMP_ENTRY_SIZE;
> cq_spec.modr_ctx_id = 0;
> - eq = &mpc->ac->eqs[cq->comp_vector];
> + eq = &mpc->eqs[cq->comp_vector % mpc->num_queues];
^^^^^^^
> cq_spec.attached_eq = eq->eq->id;
Can mpc->eqs be NULL here? Looking at the initialization paths:
Before this patch, mpc->ac->eqs was allocated during mana_probe() and was
always non-NULL after driver initialization.
After this patch, mpc->eqs is only allocated in two places:
1. mana_alloc_queues() - called when the Ethernet interface is brought up
2. mana_ib_cfg_vport() - called when RDMA RAW QPs configure the vport
However, mana_ib_create_qp_rss() doesn't call mana_ib_cfg_vport(). If an
RSS QP is created via RDMA verbs before the Ethernet port is brought up
and before any RAW QP has configured the vport, mpc->eqs would still be
NULL.
Computing &mpc->eqs[cq->comp_vector % mpc->num_queues] from a NULL pointer
produces an invalid address, and the subsequent dereference eq->eq->id
would crash the kernel.
Should there be a NULL check here, or should mana_ib_create_qp_rss()
ensure EQs are created before reaching this point?
>
> ret = mana_create_wq_obj(mpc, mpc->port_handle, GDMA_RQ,
[ ... ]
^ permalink raw reply
* RE: [EXTERNAL] Re: [v2,net-next,1/6] net: mana: Create separate EQs for each vPort
From: Long Li @ 2026-03-06 20:45 UTC (permalink / raw)
To: Simon Horman
Cc: davem@davemloft.net, edumazet@google.com, Haiyang Zhang,
linux-rdma@vger.kernel.org, linux-hyperv@vger.kernel.org,
kuba@kernel.org, wei.liu@kernel.org,
shradhagupta@linux.microsoft.com, pabeni@redhat.com,
Konstantin Taranov, schakrabarti@linux.microsoft.com,
erick.archer@outlook.com, linux-kernel@vger.kernel.org,
KY Srinivasan, Dexuan Cui, netdev@vger.kernel.org
In-Reply-To: <20260306160820.525640-1-horms@kernel.org>
> Subject: [EXTERNAL] Re: [v2,net-next,1/6] net: mana: Create separate EQs for
> each vPort
>
> This is an AI-generated review of your patch. The human sending this email has
> considered the AI review valid, or at least plausible.
>
> For local reproduction steps see
> https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fnetdev-
> ai.bots.linux.dev%2Fai-
> local.html&data=05%7C02%7Clongli%40microsoft.com%7C070979f149574e63
> e15a08de7b9aa19e%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C6
> 39084101208945710%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnR
> ydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3
> D%3D%7C0%7C%7C%7C&sdata=Cs%2BZwNguYa8FGbpb4H0naVDhDzhjzBvDU7
> 5aZekerzc%3D&reserved=0
> ---
> net: mana: Create separate EQs for each vPort
>
> To prepare for assigning vPorts to dedicated MSI-X vectors, this commit
> removes EQ sharing among vPorts and creates dedicated EQs for each vPort.
> The EQ definition is moved from struct mana_context to struct
> mana_port_context and related support functions are updated.
>
> > diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h
> > --- a/include/net/mana/mana.h
> > +++ b/include/net/mana/mana.h
>
> [ ... ]
>
> > @@ -499,6 +497,9 @@ struct mana_port_context {
> >
> > u8 mac_addr[ETH_ALEN];
> >
> > + struct mana_eq *eqs;
> > + struct dentry *mana_eqs_debugfs;
> > +
> > enum TRI_STATE rss_state;
> >
> > mana_handle_t default_rxobj;
>
> [ ... ]
>
> > diff --git a/drivers/infiniband/hw/mana/qp.c
> > b/drivers/infiniband/hw/mana/qp.c
> > --- a/drivers/infiniband/hw/mana/qp.c
> > +++ b/drivers/infiniband/hw/mana/qp.c
> > @@ -188,7 +188,7 @@ struct mana_obj_spec cq_spec = {};
> > cq_spec.gdma_region = cq->queue.gdma_region;
> > cq_spec.queue_size = cq->cqe * COMP_ENTRY_SIZE;
> > cq_spec.modr_ctx_id = 0;
> > - eq = &mpc->ac->eqs[cq->comp_vector];
> > + eq = &mpc->eqs[cq->comp_vector % mpc->num_queues];
> ^^^^^^^
> > cq_spec.attached_eq = eq->eq->id;
>
> Can mpc->eqs be NULL here? Looking at the initialization paths:
>
> Before this patch, mpc->ac->eqs was allocated during mana_probe() and was
> always non-NULL after driver initialization.
>
> After this patch, mpc->eqs is only allocated in two places:
> 1. mana_alloc_queues() - called when the Ethernet interface is brought up 2.
> mana_ib_cfg_vport() - called when RDMA RAW QPs configure the vport
>
> However, mana_ib_create_qp_rss() doesn't call mana_ib_cfg_vport(). If an RSS
> QP is created via RDMA verbs before the Ethernet port is brought up and
> before any RAW QP has configured the vport, mpc->eqs would still be NULL.
>
> Computing &mpc->eqs[cq->comp_vector % mpc->num_queues] from a NULL
> pointer produces an invalid address, and the subsequent dereference eq->eq-
> >id would crash the kernel.
>
> Should there be a NULL check here, or should mana_ib_create_qp_rss() ensure
> EQs are created before reaching this point?
I'm fixing it in v3.
Thank you,
Long
^ permalink raw reply
* [PATCH net-next] net: mana: hardening: Validate doorbell ID from GDMA_REGISTER_DEVICE response
From: Erni Sri Satya Vennela @ 2026-03-06 21:12 UTC (permalink / raw)
To: kys, haiyangz, wei.liu, decui, longli, andrew+netdev, davem,
edumazet, kuba, pabeni, kotaranov, horms, shradhagupta,
dipayanroy, yury.norov, kees, linux-hyperv, netdev, linux-kernel,
linux-rdma
Cc: Erni Sri Satya Vennela
As a part of MANA hardening for CVM, add validation for the doorbell
ID (db_id) received from hardware in the GDMA_REGISTER_DEVICE response
to prevent out-of-bounds memory access when calculating the doorbell
page address.
In mana_gd_ring_doorbell(), the doorbell page address is calculated as:
addr = db_page_base + db_page_size * db_index
= (bar0_va + db_page_off) + db_page_size * db_index
A hardware could return values that cause this address to fall outside
the BAR0 MMIO region. In Confidential VM environments, hardware responses
cannot be fully trusted.
Add the following validations:
- Store the BAR0 size (bar0_size) in gdma_context during probe.
- Validate the doorbell page offset (db_page_off) read from device
registers does not exceed bar0_size during initialization, converting
mana_gd_init_registers() to return an error code.
- Validate db_id from GDMA_REGISTER_DEVICE response against the
maximum number of doorbell pages that fit within BAR0.
Signed-off-by: Erni Sri Satya Vennela <ernis@linux.microsoft.com>
---
.../net/ethernet/microsoft/mana/gdma_main.c | 60 ++++++++++++++-----
include/net/mana/gdma.h | 4 +-
2 files changed, 49 insertions(+), 15 deletions(-)
diff --git a/drivers/net/ethernet/microsoft/mana/gdma_main.c b/drivers/net/ethernet/microsoft/mana/gdma_main.c
index aef8612b73cb..ef0dbfaac8f4 100644
--- a/drivers/net/ethernet/microsoft/mana/gdma_main.c
+++ b/drivers/net/ethernet/microsoft/mana/gdma_main.c
@@ -39,49 +39,66 @@ static u64 mana_gd_r64(struct gdma_context *g, u64 offset)
return readq(g->bar0_va + offset);
}
-static void mana_gd_init_pf_regs(struct pci_dev *pdev)
+static int mana_gd_init_pf_regs(struct pci_dev *pdev)
{
struct gdma_context *gc = pci_get_drvdata(pdev);
void __iomem *sriov_base_va;
u64 sriov_base_off;
gc->db_page_size = mana_gd_r32(gc, GDMA_PF_REG_DB_PAGE_SIZE) & 0xFFFF;
- gc->db_page_base = gc->bar0_va +
- mana_gd_r64(gc, GDMA_PF_REG_DB_PAGE_OFF);
+ gc->db_page_off = mana_gd_r64(gc, GDMA_PF_REG_DB_PAGE_OFF);
- gc->phys_db_page_base = gc->bar0_pa +
- mana_gd_r64(gc, GDMA_PF_REG_DB_PAGE_OFF);
+ /* Validate doorbell offset is within BAR0 */
+ if (gc->db_page_off >= gc->bar0_size) {
+ dev_err(gc->dev,
+ "Doorbell offset 0x%llx exceeds BAR0 size 0x%llx\n",
+ gc->db_page_off, (u64)gc->bar0_size);
+ return -EPROTO;
+ }
+
+ gc->db_page_base = gc->bar0_va + gc->db_page_off;
+ gc->phys_db_page_base = gc->bar0_pa + gc->db_page_off;
sriov_base_off = mana_gd_r64(gc, GDMA_SRIOV_REG_CFG_BASE_OFF);
sriov_base_va = gc->bar0_va + sriov_base_off;
gc->shm_base = sriov_base_va +
mana_gd_r64(gc, sriov_base_off + GDMA_PF_REG_SHM_OFF);
+
+ return 0;
}
-static void mana_gd_init_vf_regs(struct pci_dev *pdev)
+static int mana_gd_init_vf_regs(struct pci_dev *pdev)
{
struct gdma_context *gc = pci_get_drvdata(pdev);
gc->db_page_size = mana_gd_r32(gc, GDMA_REG_DB_PAGE_SIZE) & 0xFFFF;
+ gc->db_page_off = mana_gd_r64(gc, GDMA_REG_DB_PAGE_OFFSET);
- gc->db_page_base = gc->bar0_va +
- mana_gd_r64(gc, GDMA_REG_DB_PAGE_OFFSET);
+ /* Validate doorbell offset is within BAR0 */
+ if (gc->db_page_off >= gc->bar0_size) {
+ dev_err(gc->dev,
+ "Doorbell offset 0x%llx exceeds BAR0 size 0x%llx\n",
+ gc->db_page_off, (u64)gc->bar0_size);
+ return -EPROTO;
+ }
- gc->phys_db_page_base = gc->bar0_pa +
- mana_gd_r64(gc, GDMA_REG_DB_PAGE_OFFSET);
+ gc->db_page_base = gc->bar0_va + gc->db_page_off;
+ gc->phys_db_page_base = gc->bar0_pa + gc->db_page_off;
gc->shm_base = gc->bar0_va + mana_gd_r64(gc, GDMA_REG_SHM_OFFSET);
+
+ return 0;
}
-static void mana_gd_init_registers(struct pci_dev *pdev)
+static int mana_gd_init_registers(struct pci_dev *pdev)
{
struct gdma_context *gc = pci_get_drvdata(pdev);
if (gc->is_pf)
- mana_gd_init_pf_regs(pdev);
+ return mana_gd_init_pf_regs(pdev);
else
- mana_gd_init_vf_regs(pdev);
+ return mana_gd_init_vf_regs(pdev);
}
/* Suppress logging when we set timeout to zero */
@@ -1256,6 +1273,17 @@ int mana_gd_register_device(struct gdma_dev *gd)
return err ? err : -EPROTO;
}
+ /* Validate that doorbell page for db_id is within the BAR0 region.
+ * In mana_gd_ring_doorbell(), the address is calculated as:
+ * addr = db_page_base + db_page_size * db_id
+ * = (bar0_va + db_page_off) + (db_page_size * db_id)
+ * So we need: db_page_off + db_page_size * (db_id + 1) <= bar0_size
+ */
+ if (gc->db_page_off + gc->db_page_size * ((u64)resp.db_id + 1) > gc->bar0_size) {
+ dev_err(gc->dev, "Doorbell ID %u out of range\n", resp.db_id);
+ return -EPROTO;
+ }
+
gd->pdid = resp.pdid;
gd->gpa_mkey = resp.gpa_mkey;
gd->doorbell = resp.db_id;
@@ -1890,7 +1918,10 @@ static int mana_gd_setup(struct pci_dev *pdev)
struct gdma_context *gc = pci_get_drvdata(pdev);
int err;
- mana_gd_init_registers(pdev);
+ err = mana_gd_init_registers(pdev);
+ if (err)
+ return err;
+
mana_smc_init(&gc->shm_channel, gc->dev, gc->shm_base);
gc->service_wq = alloc_ordered_workqueue("gdma_service_wq", 0);
@@ -1996,6 +2027,7 @@ static int mana_gd_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
mutex_init(&gc->eq_test_event_mutex);
pci_set_drvdata(pdev, gc);
gc->bar0_pa = pci_resource_start(pdev, 0);
+ gc->bar0_size = pci_resource_len(pdev, 0);
bar0_va = pci_iomap(pdev, bar, 0);
if (!bar0_va)
diff --git a/include/net/mana/gdma.h b/include/net/mana/gdma.h
index ec17004b10c0..7fe3a1b61b2d 100644
--- a/include/net/mana/gdma.h
+++ b/include/net/mana/gdma.h
@@ -421,10 +421,12 @@ struct gdma_context {
phys_addr_t bar0_pa;
void __iomem *bar0_va;
+ resource_size_t bar0_size;
void __iomem *shm_base;
void __iomem *db_page_base;
phys_addr_t phys_db_page_base;
- u32 db_page_size;
+ u64 db_page_off;
+ u64 db_page_size;
int numa_node;
/* Shared memory chanenl (used to bootstrap HWC) */
--
2.34.1
^ permalink raw reply related
* [PATCH net-next v3 0/6] net: mana: Per-vPort EQ and MSI-X interrupt management
From: Long Li @ 2026-03-06 21:32 UTC (permalink / raw)
To: K . Y . Srinivasan, Haiyang Zhang, Wei Liu, Dexuan Cui,
David S . Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Shradha Gupta, Simon Horman, Konstantin Taranov,
Souradeep Chakrabarti, Erick Archer, linux-hyperv, netdev,
linux-kernel, linux-rdma
Cc: Long Li
This series adds per-vPort Event Queue (EQ) allocation and MSI-X interrupt
management for the MANA driver. Previously, all vPorts shared a single set
of EQs. This change enables dedicated EQs per vPort with support for both
dedicated and shared MSI-X vector allocation modes.
Patch 1 moves EQ ownership from mana_context to per-vPort mana_port_context
and exports create/destroy functions for the RDMA driver.
Patch 2 adds device capability queries to determine whether MSI-X vectors
should be dedicated per-vPort or shared. When the number of available MSI-X
vectors is insufficient for dedicated allocation, the driver enables sharing
mode with bitmap-based vector assignment.
Patch 3 introduces the GIC (GDMA IRQ Context) abstraction with reference
counting, allowing multiple EQs to safely share a single MSI-X vector.
Patch 4 converts the global EQ allocation in probe/resume to use the new
GIC functions.
Patch 5 adds per-vPort GIC lifecycle management, calling get/put on each
EQ creation and destruction during vPort open/close.
Patch 6 extends the same GIC lifecycle management to the RDMA driver's EQ
allocation path.
Tested on Azure VMs with both MSI-X sharing mode 0 (dedicated) and mode 1
(shared): NIC down/up tests, iperf3 traffic tests up to 181 Gbps.
Changes in v3:
- Rebased on net-next/main
- Patch 1: Added NULL check for mpc->eqs in mana_ib_create_qp_rss() to
prevent NULL pointer dereference when RSS QP is created before a raw QP
has configured the vport and allocated EQs
Changes in v2:
- Rebased on net-next/main (adapted to kzalloc_objs/kzalloc_obj macros,
new GDMA_DRV_CAP_FLAG definitions)
- Patch 2: Fixed misleading comment for max_num_queues vs
max_num_queues_vport in gdma.h
- Patch 3: Fixed spelling typo in gdma_main.c ("difference" -> "different")
Long Li (6):
net: mana: Create separate EQs for each vPort
net: mana: Query device capabilities and configure MSI-X sharing for
EQs
net: mana: Introduce GIC context with refcounting for interrupt
management
net: mana: Use GIC functions to allocate global EQs
net: mana: Allocate interrupt context for each EQ when creating vPort
RDMA/mana_ib: Allocate interrupt contexts on EQs
drivers/infiniband/hw/mana/main.c | 47 ++-
drivers/infiniband/hw/mana/qp.c | 12 +-
.../net/ethernet/microsoft/mana/gdma_main.c | 309 +++++++++++++-----
drivers/net/ethernet/microsoft/mana/mana_en.c | 162 +++++----
include/net/mana/gdma.h | 31 +-
include/net/mana/mana.h | 7 +-
6 files changed, 410 insertions(+), 158 deletions(-)
--
2.43.0
^ permalink raw reply
* [PATCH net-next v3 1/6] net: mana: Create separate EQs for each vPort
From: Long Li @ 2026-03-06 21:32 UTC (permalink / raw)
To: K . Y . Srinivasan, Haiyang Zhang, Wei Liu, Dexuan Cui,
David S . Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Shradha Gupta, Simon Horman, Konstantin Taranov,
Souradeep Chakrabarti, Erick Archer, linux-hyperv, netdev,
linux-kernel, linux-rdma
Cc: Long Li
In-Reply-To: <20260306213302.544681-1-longli@microsoft.com>
To prepare for assigning vPorts to dedicated MSI-X vectors, remove EQ
sharing among the vPorts and create dedicated EQs for each vPort.
Move the EQ definition from struct mana_context to struct mana_port_context
and update related support functions. Export mana_create_eq() and
mana_destroy_eq() for use by the MANA RDMA driver.
Signed-off-by: Long Li <longli@microsoft.com>
---
Changes in v3:
- Added NULL check for mpc->eqs in mana_ib_create_qp_rss() to prevent
kernel crash when RSS QP is created before EQs are allocated
---
drivers/infiniband/hw/mana/main.c | 14 ++-
drivers/infiniband/hw/mana/qp.c | 12 +-
drivers/net/ethernet/microsoft/mana/mana_en.c | 109 ++++++++++--------
include/net/mana/mana.h | 7 +-
4 files changed, 90 insertions(+), 52 deletions(-)
diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c
index 8d99cd00f002..d51dd0ee85f4 100644
--- a/drivers/infiniband/hw/mana/main.c
+++ b/drivers/infiniband/hw/mana/main.c
@@ -20,8 +20,10 @@ void mana_ib_uncfg_vport(struct mana_ib_dev *dev, struct mana_ib_pd *pd,
pd->vport_use_count--;
WARN_ON(pd->vport_use_count < 0);
- if (!pd->vport_use_count)
+ if (!pd->vport_use_count) {
+ mana_destroy_eq(mpc);
mana_uncfg_vport(mpc);
+ }
mutex_unlock(&pd->vport_mutex);
}
@@ -55,15 +57,21 @@ int mana_ib_cfg_vport(struct mana_ib_dev *dev, u32 port, struct mana_ib_pd *pd,
return err;
}
- mutex_unlock(&pd->vport_mutex);
pd->tx_shortform_allowed = mpc->tx_shortform_allowed;
pd->tx_vp_offset = mpc->tx_vp_offset;
+ err = mana_create_eq(mpc);
+ if (err) {
+ mana_uncfg_vport(mpc);
+ pd->vport_use_count--;
+ }
+
+ mutex_unlock(&pd->vport_mutex);
ibdev_dbg(&dev->ib_dev, "vport handle %llx pdid %x doorbell_id %x\n",
mpc->port_handle, pd->pdn, doorbell_id);
- return 0;
+ return err;
}
int mana_ib_alloc_pd(struct ib_pd *ibpd, struct ib_udata *udata)
diff --git a/drivers/infiniband/hw/mana/qp.c b/drivers/infiniband/hw/mana/qp.c
index 82f84f7ad37a..039c86eca0dc 100644
--- a/drivers/infiniband/hw/mana/qp.c
+++ b/drivers/infiniband/hw/mana/qp.c
@@ -188,7 +188,15 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
cq_spec.gdma_region = cq->queue.gdma_region;
cq_spec.queue_size = cq->cqe * COMP_ENTRY_SIZE;
cq_spec.modr_ctx_id = 0;
- eq = &mpc->ac->eqs[cq->comp_vector];
+ /* EQs are created when a raw QP configures the vport.
+ * A raw QP must be created before creating rwq_ind_tbl.
+ */
+ if (!mpc->eqs) {
+ ret = -EINVAL;
+ i--;
+ goto fail;
+ }
+ eq = &mpc->eqs[cq->comp_vector % mpc->num_queues];
cq_spec.attached_eq = eq->eq->id;
ret = mana_create_wq_obj(mpc, mpc->port_handle, GDMA_RQ,
@@ -340,7 +348,7 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
cq_spec.queue_size = send_cq->cqe * COMP_ENTRY_SIZE;
cq_spec.modr_ctx_id = 0;
eq_vec = send_cq->comp_vector;
- eq = &mpc->ac->eqs[eq_vec];
+ eq = &mpc->eqs[eq_vec % mpc->num_queues];
cq_spec.attached_eq = eq->eq->id;
err = mana_create_wq_obj(mpc, mpc->port_handle, GDMA_SQ, &wq_spec,
diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
index 56ee993e3a43..428dafaf315b 100644
--- a/drivers/net/ethernet/microsoft/mana/mana_en.c
+++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
@@ -1588,78 +1588,82 @@ void mana_destroy_wq_obj(struct mana_port_context *apc, u32 wq_type,
}
EXPORT_SYMBOL_NS(mana_destroy_wq_obj, "NET_MANA");
-static void mana_destroy_eq(struct mana_context *ac)
+void mana_destroy_eq(struct mana_port_context *apc)
{
+ struct mana_context *ac = apc->ac;
struct gdma_context *gc = ac->gdma_dev->gdma_context;
struct gdma_queue *eq;
int i;
- if (!ac->eqs)
+ if (!apc->eqs)
return;
- debugfs_remove_recursive(ac->mana_eqs_debugfs);
- ac->mana_eqs_debugfs = NULL;
+ debugfs_remove_recursive(apc->mana_eqs_debugfs);
+ apc->mana_eqs_debugfs = NULL;
- for (i = 0; i < gc->max_num_queues; i++) {
- eq = ac->eqs[i].eq;
+ for (i = 0; i < apc->num_queues; i++) {
+ eq = apc->eqs[i].eq;
if (!eq)
continue;
mana_gd_destroy_queue(gc, eq);
}
- kfree(ac->eqs);
- ac->eqs = NULL;
+ kfree(apc->eqs);
+ apc->eqs = NULL;
}
+EXPORT_SYMBOL_NS(mana_destroy_eq, "NET_MANA");
-static void mana_create_eq_debugfs(struct mana_context *ac, int i)
+static void mana_create_eq_debugfs(struct mana_port_context *apc, int i)
{
- struct mana_eq eq = ac->eqs[i];
+ struct mana_eq eq = apc->eqs[i];
char eqnum[32];
sprintf(eqnum, "eq%d", i);
- eq.mana_eq_debugfs = debugfs_create_dir(eqnum, ac->mana_eqs_debugfs);
+ eq.mana_eq_debugfs = debugfs_create_dir(eqnum, apc->mana_eqs_debugfs);
debugfs_create_u32("head", 0400, eq.mana_eq_debugfs, &eq.eq->head);
debugfs_create_u32("tail", 0400, eq.mana_eq_debugfs, &eq.eq->tail);
debugfs_create_file("eq_dump", 0400, eq.mana_eq_debugfs, eq.eq, &mana_dbg_q_fops);
}
-static int mana_create_eq(struct mana_context *ac)
+int mana_create_eq(struct mana_port_context *apc)
{
- struct gdma_dev *gd = ac->gdma_dev;
+ struct gdma_dev *gd = apc->ac->gdma_dev;
struct gdma_context *gc = gd->gdma_context;
struct gdma_queue_spec spec = {};
int err;
int i;
- ac->eqs = kzalloc_objs(struct mana_eq, gc->max_num_queues);
- if (!ac->eqs)
+ WARN_ON(apc->eqs);
+ apc->eqs = kzalloc_objs(struct mana_eq, apc->num_queues);
+ if (!apc->eqs)
return -ENOMEM;
spec.type = GDMA_EQ;
spec.monitor_avl_buf = false;
spec.queue_size = EQ_SIZE;
spec.eq.callback = NULL;
- spec.eq.context = ac->eqs;
+ spec.eq.context = apc->eqs;
spec.eq.log2_throttle_limit = LOG2_EQ_THROTTLE;
- ac->mana_eqs_debugfs = debugfs_create_dir("EQs", gc->mana_pci_debugfs);
+ apc->mana_eqs_debugfs = debugfs_create_dir("EQs", apc->mana_port_debugfs);
- for (i = 0; i < gc->max_num_queues; i++) {
+ for (i = 0; i < apc->num_queues; i++) {
spec.eq.msix_index = (i + 1) % gc->num_msix_usable;
- err = mana_gd_create_mana_eq(gd, &spec, &ac->eqs[i].eq);
+ err = mana_gd_create_mana_eq(gd, &spec, &apc->eqs[i].eq);
if (err) {
dev_err(gc->dev, "Failed to create EQ %d : %d\n", i, err);
goto out;
}
- mana_create_eq_debugfs(ac, i);
+ mana_create_eq_debugfs(apc, i);
}
return 0;
out:
- mana_destroy_eq(ac);
+ mana_destroy_eq(apc);
return err;
}
+EXPORT_SYMBOL_NS(mana_create_eq, "NET_MANA");
static int mana_fence_rq(struct mana_port_context *apc, struct mana_rxq *rxq)
{
@@ -2377,7 +2381,7 @@ static int mana_create_txq(struct mana_port_context *apc,
spec.monitor_avl_buf = false;
spec.queue_size = cq_size;
spec.cq.callback = mana_schedule_napi;
- spec.cq.parent_eq = ac->eqs[i].eq;
+ spec.cq.parent_eq = apc->eqs[i].eq;
spec.cq.context = cq;
err = mana_gd_create_mana_wq_cq(gd, &spec, &cq->gdma_cq);
if (err)
@@ -2770,13 +2774,12 @@ static void mana_create_rxq_debugfs(struct mana_port_context *apc, int idx)
static int mana_add_rx_queues(struct mana_port_context *apc,
struct net_device *ndev)
{
- struct mana_context *ac = apc->ac;
struct mana_rxq *rxq;
int err = 0;
int i;
for (i = 0; i < apc->num_queues; i++) {
- rxq = mana_create_rxq(apc, i, &ac->eqs[i], ndev);
+ rxq = mana_create_rxq(apc, i, &apc->eqs[i], ndev);
if (!rxq) {
err = -ENOMEM;
netdev_err(ndev, "Failed to create rxq %d : %d\n", i, err);
@@ -2795,9 +2798,8 @@ static int mana_add_rx_queues(struct mana_port_context *apc,
return err;
}
-static void mana_destroy_vport(struct mana_port_context *apc)
+static void mana_destroy_rxqs(struct mana_port_context *apc)
{
- struct gdma_dev *gd = apc->ac->gdma_dev;
struct mana_rxq *rxq;
u32 rxq_idx;
@@ -2809,8 +2811,12 @@ static void mana_destroy_vport(struct mana_port_context *apc)
mana_destroy_rxq(apc, rxq, true);
apc->rxqs[rxq_idx] = NULL;
}
+}
+
+static void mana_destroy_vport(struct mana_port_context *apc)
+{
+ struct gdma_dev *gd = apc->ac->gdma_dev;
- mana_destroy_txq(apc);
mana_uncfg_vport(apc);
if (gd->gdma_context->is_pf && !apc->ac->bm_hostmode)
@@ -2831,11 +2837,7 @@ static int mana_create_vport(struct mana_port_context *apc,
return err;
}
- err = mana_cfg_vport(apc, gd->pdid, gd->doorbell);
- if (err)
- return err;
-
- return mana_create_txq(apc, net);
+ return mana_cfg_vport(apc, gd->pdid, gd->doorbell);
}
static int mana_rss_table_alloc(struct mana_port_context *apc)
@@ -3112,21 +3114,36 @@ int mana_alloc_queues(struct net_device *ndev)
err = mana_create_vport(apc, ndev);
if (err) {
- netdev_err(ndev, "Failed to create vPort %u : %d\n", apc->port_idx, err);
+ netdev_err(ndev, "Failed to create vPort %u : %d\n",
+ apc->port_idx, err);
return err;
}
+ err = mana_create_eq(apc);
+ if (err) {
+ netdev_err(ndev, "Failed to create EQ on vPort %u: %d\n",
+ apc->port_idx, err);
+ goto destroy_vport;
+ }
+
+ err = mana_create_txq(apc, ndev);
+ if (err) {
+ netdev_err(ndev, "Failed to create TXQ on vPort %u: %d\n",
+ apc->port_idx, err);
+ goto destroy_eq;
+ }
+
err = netif_set_real_num_tx_queues(ndev, apc->num_queues);
if (err) {
netdev_err(ndev,
"netif_set_real_num_tx_queues () failed for ndev with num_queues %u : %d\n",
apc->num_queues, err);
- goto destroy_vport;
+ goto destroy_txq;
}
err = mana_add_rx_queues(apc, ndev);
if (err)
- goto destroy_vport;
+ goto destroy_rxq;
apc->rss_state = apc->num_queues > 1 ? TRI_STATE_TRUE : TRI_STATE_FALSE;
@@ -3135,7 +3152,7 @@ int mana_alloc_queues(struct net_device *ndev)
netdev_err(ndev,
"netif_set_real_num_rx_queues () failed for ndev with num_queues %u : %d\n",
apc->num_queues, err);
- goto destroy_vport;
+ goto destroy_rxq;
}
mana_rss_table_init(apc);
@@ -3143,19 +3160,25 @@ int mana_alloc_queues(struct net_device *ndev)
err = mana_config_rss(apc, TRI_STATE_TRUE, true, true);
if (err) {
netdev_err(ndev, "Failed to configure RSS table: %d\n", err);
- goto destroy_vport;
+ goto destroy_rxq;
}
if (gd->gdma_context->is_pf && !apc->ac->bm_hostmode) {
err = mana_pf_register_filter(apc);
if (err)
- goto destroy_vport;
+ goto destroy_rxq;
}
mana_chn_setxdp(apc, mana_xdp_get(apc));
return 0;
+destroy_rxq:
+ mana_destroy_rxqs(apc);
+destroy_txq:
+ mana_destroy_txq(apc);
+destroy_eq:
+ mana_destroy_eq(apc);
destroy_vport:
mana_destroy_vport(apc);
return err;
@@ -3258,6 +3281,9 @@ static int mana_dealloc_queues(struct net_device *ndev)
netdev_err(ndev, "Failed to disable vPort: %d\n", err);
/* Even in err case, still need to cleanup the vPort */
+ mana_destroy_rxqs(apc);
+ mana_destroy_txq(apc);
+ mana_destroy_eq(apc);
mana_destroy_vport(apc);
return 0;
@@ -3572,12 +3598,6 @@ int mana_probe(struct gdma_dev *gd, bool resuming)
gd->driver_data = ac;
}
- err = mana_create_eq(ac);
- if (err) {
- dev_err(dev, "Failed to create EQs: %d\n", err);
- goto out;
- }
-
err = mana_query_device_cfg(ac, MANA_MAJOR_VERSION, MANA_MINOR_VERSION,
MANA_MICRO_VERSION, &num_ports, &bm_hostmode);
if (err)
@@ -3716,7 +3736,6 @@ void mana_remove(struct gdma_dev *gd, bool suspending)
free_netdev(ndev);
}
- mana_destroy_eq(ac);
out:
if (ac->per_port_queue_reset_wq) {
destroy_workqueue(ac->per_port_queue_reset_wq);
diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h
index a078af283bdd..787e637059df 100644
--- a/include/net/mana/mana.h
+++ b/include/net/mana/mana.h
@@ -478,8 +478,6 @@ struct mana_context {
u8 bm_hostmode;
struct mana_ethtool_hc_stats hc_stats;
- struct mana_eq *eqs;
- struct dentry *mana_eqs_debugfs;
struct workqueue_struct *per_port_queue_reset_wq;
/* Workqueue for querying hardware stats */
struct delayed_work gf_stats_work;
@@ -499,6 +497,9 @@ struct mana_port_context {
u8 mac_addr[ETH_ALEN];
+ struct mana_eq *eqs;
+ struct dentry *mana_eqs_debugfs;
+
enum TRI_STATE rss_state;
mana_handle_t default_rxobj;
@@ -1023,6 +1024,8 @@ void mana_destroy_wq_obj(struct mana_port_context *apc, u32 wq_type,
int mana_cfg_vport(struct mana_port_context *apc, u32 protection_dom_id,
u32 doorbell_pg_id);
void mana_uncfg_vport(struct mana_port_context *apc);
+int mana_create_eq(struct mana_port_context *apc);
+void mana_destroy_eq(struct mana_port_context *apc);
struct net_device *mana_get_primary_netdev(struct mana_context *ac,
u32 port_index,
--
2.43.0
^ permalink raw reply related
* [PATCH net-next v3 2/6] net: mana: Query device capabilities and configure MSI-X sharing for EQs
From: Long Li @ 2026-03-06 21:32 UTC (permalink / raw)
To: K . Y . Srinivasan, Haiyang Zhang, Wei Liu, Dexuan Cui,
David S . Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Shradha Gupta, Simon Horman, Konstantin Taranov,
Souradeep Chakrabarti, Erick Archer, linux-hyperv, netdev,
linux-kernel, linux-rdma
Cc: Long Li
In-Reply-To: <20260306213302.544681-1-longli@microsoft.com>
When querying the device, adjust the max number of queues to allow
dedicated MSI-X vectors for each vPort. The number of queues per vPort
is clamped to no less than 16. MSI-X sharing among vPorts is disabled
by default and is only enabled when there are not enough MSI-X vectors
for dedicated allocation.
Rename mana_query_device_cfg() to mana_gd_query_device_cfg() as it is
used at GDMA device probe time for querying device capabilities.
Signed-off-by: Long Li <longli@microsoft.com>
---
.../net/ethernet/microsoft/mana/gdma_main.c | 66 ++++++++++++++++---
drivers/net/ethernet/microsoft/mana/mana_en.c | 36 +++++-----
include/net/mana/gdma.h | 13 +++-
3 files changed, 91 insertions(+), 24 deletions(-)
diff --git a/drivers/net/ethernet/microsoft/mana/gdma_main.c b/drivers/net/ethernet/microsoft/mana/gdma_main.c
index aef8612b73cb..a6ab2f053fe9 100644
--- a/drivers/net/ethernet/microsoft/mana/gdma_main.c
+++ b/drivers/net/ethernet/microsoft/mana/gdma_main.c
@@ -107,6 +107,9 @@ static int mana_gd_query_max_resources(struct pci_dev *pdev)
struct gdma_context *gc = pci_get_drvdata(pdev);
struct gdma_query_max_resources_resp resp = {};
struct gdma_general_req req = {};
+ unsigned int max_num_queues;
+ u8 bm_hostmode;
+ u16 num_ports;
int err;
mana_gd_init_req_hdr(&req.hdr, GDMA_QUERY_MAX_RESOURCES,
@@ -152,6 +155,40 @@ static int mana_gd_query_max_resources(struct pci_dev *pdev)
if (gc->max_num_queues > gc->num_msix_usable - 1)
gc->max_num_queues = gc->num_msix_usable - 1;
+ err = mana_gd_query_device_cfg(gc, MANA_MAJOR_VERSION, MANA_MINOR_VERSION,
+ MANA_MICRO_VERSION, &num_ports, &bm_hostmode);
+ if (err)
+ return err;
+
+ if (!num_ports)
+ return -EINVAL;
+
+ /*
+ * Adjust gc->max_num_queues returned from the SOC to allow dedicated MSIx
+ * for each vPort. Reduce max_num_queues to no less than 16 if necessary
+ */
+ max_num_queues = (gc->num_msix_usable - 1) / num_ports;
+ max_num_queues = roundup_pow_of_two(max(max_num_queues, 1U));
+ if (max_num_queues < 16)
+ max_num_queues = 16;
+
+ /*
+ * Use dedicated MSIx for EQs whenever possible, use MSIx sharing for
+ * Ethernet EQs when (max_num_queues * num_ports > num_msix_usable - 1)
+ */
+ max_num_queues = min(gc->max_num_queues, max_num_queues);
+ if (max_num_queues * num_ports > gc->num_msix_usable - 1)
+ gc->msi_sharing = true;
+
+ /* If MSI is shared, use max allowed value */
+ if (gc->msi_sharing)
+ gc->max_num_queues_vport = min(gc->num_msix_usable - 1, gc->max_num_queues);
+ else
+ gc->max_num_queues_vport = max_num_queues;
+
+ dev_info(gc->dev, "MSI sharing mode %d max queues %d\n",
+ gc->msi_sharing, gc->max_num_queues);
+
return 0;
}
@@ -1803,6 +1840,7 @@ static int mana_gd_setup_hwc_irqs(struct pci_dev *pdev)
/* Need 1 interrupt for HWC */
max_irqs = min(num_online_cpus(), MANA_MAX_NUM_QUEUES) + 1;
min_irqs = 2;
+ gc->msi_sharing = true;
}
nvec = pci_alloc_irq_vectors(pdev, min_irqs, max_irqs, PCI_IRQ_MSIX);
@@ -1881,6 +1919,8 @@ static void mana_gd_remove_irqs(struct pci_dev *pdev)
pci_free_irq_vectors(pdev);
+ bitmap_free(gc->msi_bitmap);
+ gc->msi_bitmap = NULL;
gc->max_num_msix = 0;
gc->num_msix_usable = 0;
}
@@ -1912,20 +1952,30 @@ static int mana_gd_setup(struct pci_dev *pdev)
if (err)
goto destroy_hwc;
- err = mana_gd_query_max_resources(pdev);
+ err = mana_gd_detect_devices(pdev);
if (err)
goto destroy_hwc;
- err = mana_gd_setup_remaining_irqs(pdev);
- if (err) {
- dev_err(gc->dev, "Failed to setup remaining IRQs: %d", err);
- goto destroy_hwc;
- }
-
- err = mana_gd_detect_devices(pdev);
+ err = mana_gd_query_max_resources(pdev);
if (err)
goto destroy_hwc;
+ if (!gc->msi_sharing) {
+ gc->msi_bitmap = bitmap_zalloc(gc->num_msix_usable, GFP_KERNEL);
+ if (!gc->msi_bitmap) {
+ err = -ENOMEM;
+ goto destroy_hwc;
+ }
+ /* Set bit for HWC */
+ set_bit(0, gc->msi_bitmap);
+ } else {
+ err = mana_gd_setup_remaining_irqs(pdev);
+ if (err) {
+ dev_err(gc->dev, "Failed to setup remaining IRQs: %d", err);
+ goto destroy_hwc;
+ }
+ }
+
dev_dbg(&pdev->dev, "mana gdma setup successful\n");
return 0;
diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
index 428dafaf315b..bfa0f354355d 100644
--- a/drivers/net/ethernet/microsoft/mana/mana_en.c
+++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
@@ -1000,10 +1000,9 @@ static int mana_init_port_context(struct mana_port_context *apc)
return !apc->rxqs ? -ENOMEM : 0;
}
-static int mana_send_request(struct mana_context *ac, void *in_buf,
- u32 in_len, void *out_buf, u32 out_len)
+static int gdma_mana_send_request(struct gdma_context *gc, void *in_buf,
+ u32 in_len, void *out_buf, u32 out_len)
{
- struct gdma_context *gc = ac->gdma_dev->gdma_context;
struct gdma_resp_hdr *resp = out_buf;
struct gdma_req_hdr *req = in_buf;
struct device *dev = gc->dev;
@@ -1037,6 +1036,14 @@ static int mana_send_request(struct mana_context *ac, void *in_buf,
return 0;
}
+static int mana_send_request(struct mana_context *ac, void *in_buf,
+ u32 in_len, void *out_buf, u32 out_len)
+{
+ struct gdma_context *gc = ac->gdma_dev->gdma_context;
+
+ return gdma_mana_send_request(gc, in_buf, in_len, out_buf, out_len);
+}
+
static int mana_verify_resp_hdr(const struct gdma_resp_hdr *resp_hdr,
const enum mana_command_code expected_code,
const u32 min_size)
@@ -1170,11 +1177,10 @@ static void mana_pf_deregister_filter(struct mana_port_context *apc)
err, resp.hdr.status);
}
-static int mana_query_device_cfg(struct mana_context *ac, u32 proto_major_ver,
- u32 proto_minor_ver, u32 proto_micro_ver,
- u16 *max_num_vports, u8 *bm_hostmode)
+int mana_gd_query_device_cfg(struct gdma_context *gc, u32 proto_major_ver,
+ u32 proto_minor_ver, u32 proto_micro_ver,
+ u16 *max_num_vports, u8 *bm_hostmode)
{
- struct gdma_context *gc = ac->gdma_dev->gdma_context;
struct mana_query_device_cfg_resp resp = {};
struct mana_query_device_cfg_req req = {};
struct device *dev = gc->dev;
@@ -1189,7 +1195,7 @@ static int mana_query_device_cfg(struct mana_context *ac, u32 proto_major_ver,
req.proto_minor_ver = proto_minor_ver;
req.proto_micro_ver = proto_micro_ver;
- err = mana_send_request(ac, &req, sizeof(req), &resp, sizeof(resp));
+ err = gdma_mana_send_request(gc, &req, sizeof(req), &resp, sizeof(resp));
if (err) {
dev_err(dev, "Failed to query config: %d", err);
return err;
@@ -1217,8 +1223,6 @@ static int mana_query_device_cfg(struct mana_context *ac, u32 proto_major_ver,
else
*bm_hostmode = 0;
- debugfs_create_u16("adapter-MTU", 0400, gc->mana_pci_debugfs, &gc->adapter_mtu);
-
return 0;
}
@@ -3329,7 +3333,7 @@ static int mana_probe_port(struct mana_context *ac, int port_idx,
int err;
ndev = alloc_etherdev_mq(sizeof(struct mana_port_context),
- gc->max_num_queues);
+ gc->max_num_queues_vport);
if (!ndev)
return -ENOMEM;
@@ -3338,8 +3342,8 @@ static int mana_probe_port(struct mana_context *ac, int port_idx,
apc = netdev_priv(ndev);
apc->ac = ac;
apc->ndev = ndev;
- apc->max_queues = gc->max_num_queues;
- apc->num_queues = gc->max_num_queues;
+ apc->max_queues = gc->max_num_queues_vport;
+ apc->num_queues = gc->max_num_queues_vport;
apc->tx_queue_size = DEF_TX_BUFFERS_PER_QUEUE;
apc->rx_queue_size = DEF_RX_BUFFERS_PER_QUEUE;
apc->port_handle = INVALID_MANA_HANDLE;
@@ -3598,13 +3602,15 @@ int mana_probe(struct gdma_dev *gd, bool resuming)
gd->driver_data = ac;
}
- err = mana_query_device_cfg(ac, MANA_MAJOR_VERSION, MANA_MINOR_VERSION,
- MANA_MICRO_VERSION, &num_ports, &bm_hostmode);
+ err = mana_gd_query_device_cfg(gc, MANA_MAJOR_VERSION, MANA_MINOR_VERSION,
+ MANA_MICRO_VERSION, &num_ports, &bm_hostmode);
if (err)
goto out;
ac->bm_hostmode = bm_hostmode;
+ debugfs_create_u16("adapter-MTU", 0400, gc->mana_pci_debugfs, &gc->adapter_mtu);
+
if (!resuming) {
ac->num_ports = num_ports;
diff --git a/include/net/mana/gdma.h b/include/net/mana/gdma.h
index ec17004b10c0..b744253b44e8 100644
--- a/include/net/mana/gdma.h
+++ b/include/net/mana/gdma.h
@@ -399,8 +399,10 @@ struct gdma_context {
struct device *dev;
struct dentry *mana_pci_debugfs;
- /* Per-vPort max number of queues */
+ /* Hardware max number of queues */
unsigned int max_num_queues;
+ /* Per-vPort max number of queues */
+ unsigned int max_num_queues_vport;
unsigned int max_num_msix;
unsigned int num_msix_usable;
struct xarray irq_contexts;
@@ -444,6 +446,12 @@ struct gdma_context {
struct workqueue_struct *service_wq;
unsigned long flags;
+
+ /* Indicate if this device is sharing MSI for EQs on MANA */
+ bool msi_sharing;
+
+ /* Bitmap tracks where MSI is allocated when it is not shared for EQs */
+ unsigned long *msi_bitmap;
};
static inline bool mana_gd_is_mana(struct gdma_dev *gd)
@@ -1011,4 +1019,7 @@ int mana_gd_resume(struct pci_dev *pdev);
bool mana_need_log(struct gdma_context *gc, int err);
+int mana_gd_query_device_cfg(struct gdma_context *gc, u32 proto_major_ver,
+ u32 proto_minor_ver, u32 proto_micro_ver,
+ u16 *max_num_vports, u8 *bm_hostmode);
#endif /* _GDMA_H */
--
2.43.0
^ permalink raw reply related
* [PATCH net-next v3 3/6] net: mana: Introduce GIC context with refcounting for interrupt management
From: Long Li @ 2026-03-06 21:32 UTC (permalink / raw)
To: K . Y . Srinivasan, Haiyang Zhang, Wei Liu, Dexuan Cui,
David S . Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Shradha Gupta, Simon Horman, Konstantin Taranov,
Souradeep Chakrabarti, Erick Archer, linux-hyperv, netdev,
linux-kernel, linux-rdma
Cc: Long Li
In-Reply-To: <20260306213302.544681-1-longli@microsoft.com>
To allow Ethernet EQs to use dedicated or shared MSI-X vectors and RDMA
EQs to share the same MSI-X, introduce a GIC (GDMA IRQ Context) with
reference counting. This allows the driver to create an interrupt context
on an assigned or unassigned MSI-X vector and share it across multiple
EQ consumers.
Signed-off-by: Long Li <longli@microsoft.com>
---
.../net/ethernet/microsoft/mana/gdma_main.c | 158 ++++++++++++++++++
include/net/mana/gdma.h | 10 ++
2 files changed, 168 insertions(+)
diff --git a/drivers/net/ethernet/microsoft/mana/gdma_main.c b/drivers/net/ethernet/microsoft/mana/gdma_main.c
index a6ab2f053fe9..c43fd8089e77 100644
--- a/drivers/net/ethernet/microsoft/mana/gdma_main.c
+++ b/drivers/net/ethernet/microsoft/mana/gdma_main.c
@@ -1559,6 +1559,163 @@ static irqreturn_t mana_gd_intr(int irq, void *arg)
return IRQ_HANDLED;
}
+void mana_gd_put_gic(struct gdma_context *gc, bool use_msi_bitmap, int msi)
+{
+ struct pci_dev *dev = to_pci_dev(gc->dev);
+ struct msi_map irq_map;
+ struct gdma_irq_context *gic;
+ int irq;
+
+ mutex_lock(&gc->gic_mutex);
+
+ gic = xa_load(&gc->irq_contexts, msi);
+ if (WARN_ON(!gic)) {
+ mutex_unlock(&gc->gic_mutex);
+ return;
+ }
+
+ if (use_msi_bitmap)
+ gic->bitmap_refs--;
+
+ if (use_msi_bitmap && gic->bitmap_refs == 0)
+ clear_bit(msi, gc->msi_bitmap);
+
+ if (!refcount_dec_and_test(&gic->refcount))
+ goto out;
+
+ irq = pci_irq_vector(dev, msi);
+
+ irq_update_affinity_hint(irq, NULL);
+ free_irq(irq, gic);
+
+ if (pci_msix_can_alloc_dyn(dev)) {
+ irq_map.virq = irq;
+ irq_map.index = msi;
+ pci_msix_free_irq(dev, irq_map);
+ }
+
+ xa_erase(&gc->irq_contexts, msi);
+ kfree(gic);
+
+out:
+ mutex_unlock(&gc->gic_mutex);
+}
+EXPORT_SYMBOL_NS(mana_gd_put_gic, "NET_MANA");
+
+/*
+ * Get a GIC (GDMA IRQ Context) on a MSI vector
+ * a MSI can be shared between different EQs, this function supports setting
+ * up separate MSIs using a bitmap, or directly using the MSI index
+ *
+ * @use_msi_bitmap:
+ * True if MSI is assigned by this function on available slots from bitmap.
+ * False if MSI is passed from *msi_requested
+ */
+struct gdma_irq_context *mana_gd_get_gic(struct gdma_context *gc,
+ bool use_msi_bitmap,
+ int *msi_requested)
+{
+ struct gdma_irq_context *gic;
+ struct pci_dev *dev = to_pci_dev(gc->dev);
+ struct msi_map irq_map = { };
+ int irq;
+ int msi;
+ int err;
+
+ mutex_lock(&gc->gic_mutex);
+
+ if (use_msi_bitmap) {
+ msi = find_first_zero_bit(gc->msi_bitmap, gc->num_msix_usable);
+ if (msi >= gc->num_msix_usable) {
+ dev_err(gc->dev, "No free MSI vectors available\n");
+ gic = NULL;
+ goto out;
+ }
+ *msi_requested = msi;
+ } else {
+ msi = *msi_requested;
+ }
+
+ gic = xa_load(&gc->irq_contexts, msi);
+ if (gic) {
+ refcount_inc(&gic->refcount);
+ if (use_msi_bitmap) {
+ gic->bitmap_refs++;
+ set_bit(msi, gc->msi_bitmap);
+ }
+ goto out;
+ }
+
+ irq = pci_irq_vector(dev, msi);
+ if (irq == -EINVAL) {
+ irq_map = pci_msix_alloc_irq_at(dev, msi, NULL);
+ if (!irq_map.virq) {
+ err = irq_map.index;
+ dev_err(gc->dev,
+ "Failed to alloc irq_map msi %d err %d\n",
+ msi, err);
+ gic = NULL;
+ goto out;
+ }
+ irq = irq_map.virq;
+ msi = irq_map.index;
+ }
+
+ gic = kzalloc(sizeof(*gic), GFP_KERNEL);
+ if (!gic) {
+ if (irq_map.virq)
+ pci_msix_free_irq(dev, irq_map);
+ goto out;
+ }
+
+ gic->handler = mana_gd_process_eq_events;
+ gic->msi = msi;
+ gic->irq = irq;
+ INIT_LIST_HEAD(&gic->eq_list);
+ spin_lock_init(&gic->lock);
+
+ if (!gic->msi)
+ snprintf(gic->name, MANA_IRQ_NAME_SZ, "mana_hwc@pci:%s",
+ pci_name(dev));
+ else
+ snprintf(gic->name, MANA_IRQ_NAME_SZ, "mana_msi%d@pci:%s",
+ gic->msi, pci_name(dev));
+
+ err = request_irq(irq, mana_gd_intr, 0, gic->name, gic);
+ if (err) {
+ dev_err(gc->dev, "Failed to request irq %d %s\n",
+ irq, gic->name);
+ kfree(gic);
+ gic = NULL;
+ if (irq_map.virq)
+ pci_msix_free_irq(dev, irq_map);
+ goto out;
+ }
+
+ refcount_set(&gic->refcount, 1);
+ gic->bitmap_refs = use_msi_bitmap ? 1 : 0;
+
+ err = xa_err(xa_store(&gc->irq_contexts, msi, gic, GFP_KERNEL));
+ if (err) {
+ dev_err(gc->dev, "Failed to store irq context for msi %d: %d\n",
+ msi, err);
+ free_irq(irq, gic);
+ kfree(gic);
+ gic = NULL;
+ if (irq_map.virq)
+ pci_msix_free_irq(dev, irq_map);
+ goto out;
+ }
+
+ if (use_msi_bitmap)
+ set_bit(msi, gc->msi_bitmap);
+
+out:
+ mutex_unlock(&gc->gic_mutex);
+ return gic;
+}
+EXPORT_SYMBOL_NS(mana_gd_get_gic, "NET_MANA");
+
int mana_gd_alloc_res_map(u32 res_avail, struct gdma_resource *r)
{
r->map = bitmap_zalloc(res_avail, GFP_KERNEL);
@@ -2044,6 +2201,7 @@ static int mana_gd_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
goto release_region;
mutex_init(&gc->eq_test_event_mutex);
+ mutex_init(&gc->gic_mutex);
pci_set_drvdata(pdev, gc);
gc->bar0_pa = pci_resource_start(pdev, 0);
diff --git a/include/net/mana/gdma.h b/include/net/mana/gdma.h
index b744253b44e8..1a7f4abe7a8b 100644
--- a/include/net/mana/gdma.h
+++ b/include/net/mana/gdma.h
@@ -388,6 +388,10 @@ struct gdma_irq_context {
spinlock_t lock;
struct list_head eq_list;
char name[MANA_IRQ_NAME_SZ];
+ unsigned int msi;
+ unsigned int irq;
+ refcount_t refcount;
+ unsigned int bitmap_refs;
};
enum gdma_context_flags {
@@ -447,6 +451,9 @@ struct gdma_context {
unsigned long flags;
+ /* Protect access to GIC context */
+ struct mutex gic_mutex;
+
/* Indicate if this device is sharing MSI for EQs on MANA */
bool msi_sharing;
@@ -1019,6 +1026,9 @@ int mana_gd_resume(struct pci_dev *pdev);
bool mana_need_log(struct gdma_context *gc, int err);
+struct gdma_irq_context *mana_gd_get_gic(struct gdma_context *gc, bool use_msi_bitmap,
+ int *msi_requested);
+void mana_gd_put_gic(struct gdma_context *gc, bool use_msi_bitmap, int msi);
int mana_gd_query_device_cfg(struct gdma_context *gc, u32 proto_major_ver,
u32 proto_minor_ver, u32 proto_micro_ver,
u16 *max_num_vports, u8 *bm_hostmode);
--
2.43.0
^ permalink raw reply related
* [PATCH net-next v3 4/6] net: mana: Use GIC functions to allocate global EQs
From: Long Li @ 2026-03-06 21:33 UTC (permalink / raw)
To: K . Y . Srinivasan, Haiyang Zhang, Wei Liu, Dexuan Cui,
David S . Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Shradha Gupta, Simon Horman, Konstantin Taranov,
Souradeep Chakrabarti, Erick Archer, linux-hyperv, netdev,
linux-kernel, linux-rdma
Cc: Long Li
In-Reply-To: <20260306213302.544681-1-longli@microsoft.com>
Replace the GDMA global interrupt setup code with the new GIC allocation
and release functions for managing interrupt contexts.
Signed-off-by: Long Li <longli@microsoft.com>
---
.../net/ethernet/microsoft/mana/gdma_main.c | 83 +++----------------
1 file changed, 10 insertions(+), 73 deletions(-)
diff --git a/drivers/net/ethernet/microsoft/mana/gdma_main.c b/drivers/net/ethernet/microsoft/mana/gdma_main.c
index c43fd8089e77..bdc9dc437fb7 100644
--- a/drivers/net/ethernet/microsoft/mana/gdma_main.c
+++ b/drivers/net/ethernet/microsoft/mana/gdma_main.c
@@ -1831,30 +1831,13 @@ static int mana_gd_setup_dyn_irqs(struct pci_dev *pdev, int nvec)
* further used in irq_setup()
*/
for (i = 1; i <= nvec; i++) {
- gic = kzalloc_obj(*gic);
+ gic = mana_gd_get_gic(gc, false, &i);
if (!gic) {
err = -ENOMEM;
goto free_irq;
}
- gic->handler = mana_gd_process_eq_events;
- INIT_LIST_HEAD(&gic->eq_list);
- spin_lock_init(&gic->lock);
-
- snprintf(gic->name, MANA_IRQ_NAME_SZ, "mana_q%d@pci:%s",
- i - 1, pci_name(pdev));
-
- /* one pci vector is already allocated for HWC */
- irqs[i - 1] = pci_irq_vector(pdev, i);
- if (irqs[i - 1] < 0) {
- err = irqs[i - 1];
- goto free_current_gic;
- }
-
- err = request_irq(irqs[i - 1], mana_gd_intr, 0, gic->name, gic);
- if (err)
- goto free_current_gic;
- xa_store(&gc->irq_contexts, i, gic, GFP_KERNEL);
+ irqs[i - 1] = gic->irq;
}
/*
@@ -1876,19 +1859,11 @@ static int mana_gd_setup_dyn_irqs(struct pci_dev *pdev, int nvec)
kfree(irqs);
return 0;
-free_current_gic:
- kfree(gic);
free_irq:
for (i -= 1; i > 0; i--) {
irq = pci_irq_vector(pdev, i);
- gic = xa_load(&gc->irq_contexts, i);
- if (WARN_ON(!gic))
- continue;
-
irq_update_affinity_hint(irq, NULL);
- free_irq(irq, gic);
- xa_erase(&gc->irq_contexts, i);
- kfree(gic);
+ mana_gd_put_gic(gc, false, i);
}
kfree(irqs);
return err;
@@ -1909,34 +1884,13 @@ static int mana_gd_setup_irqs(struct pci_dev *pdev, int nvec)
start_irqs = irqs;
for (i = 0; i < nvec; i++) {
- gic = kzalloc_obj(*gic);
+ gic = mana_gd_get_gic(gc, false, &i);
if (!gic) {
err = -ENOMEM;
goto free_irq;
}
- gic->handler = mana_gd_process_eq_events;
- INIT_LIST_HEAD(&gic->eq_list);
- spin_lock_init(&gic->lock);
-
- if (!i)
- snprintf(gic->name, MANA_IRQ_NAME_SZ, "mana_hwc@pci:%s",
- pci_name(pdev));
- else
- snprintf(gic->name, MANA_IRQ_NAME_SZ, "mana_q%d@pci:%s",
- i - 1, pci_name(pdev));
-
- irqs[i] = pci_irq_vector(pdev, i);
- if (irqs[i] < 0) {
- err = irqs[i];
- goto free_current_gic;
- }
-
- err = request_irq(irqs[i], mana_gd_intr, 0, gic->name, gic);
- if (err)
- goto free_current_gic;
-
- xa_store(&gc->irq_contexts, i, gic, GFP_KERNEL);
+ irqs[i] = gic->irq;
}
/* If number of IRQ is one extra than number of online CPUs,
@@ -1965,19 +1919,11 @@ static int mana_gd_setup_irqs(struct pci_dev *pdev, int nvec)
kfree(start_irqs);
return 0;
-free_current_gic:
- kfree(gic);
free_irq:
for (i -= 1; i >= 0; i--) {
irq = pci_irq_vector(pdev, i);
- gic = xa_load(&gc->irq_contexts, i);
- if (WARN_ON(!gic))
- continue;
-
irq_update_affinity_hint(irq, NULL);
- free_irq(irq, gic);
- xa_erase(&gc->irq_contexts, i);
- kfree(gic);
+ mana_gd_put_gic(gc, false, i);
}
kfree(start_irqs);
@@ -2052,26 +1998,17 @@ static int mana_gd_setup_remaining_irqs(struct pci_dev *pdev)
static void mana_gd_remove_irqs(struct pci_dev *pdev)
{
struct gdma_context *gc = pci_get_drvdata(pdev);
- struct gdma_irq_context *gic;
int irq, i;
if (gc->max_num_msix < 1)
return;
- for (i = 0; i < gc->max_num_msix; i++) {
- irq = pci_irq_vector(pdev, i);
- if (irq < 0)
- continue;
-
- gic = xa_load(&gc->irq_contexts, i);
- if (WARN_ON(!gic))
- continue;
-
+ for (i = 0; i < (gc->msi_sharing ? gc->max_num_msix : 1); i++) {
/* Need to clear the hint before free_irq */
+ irq = pci_irq_vector(pdev, i);
irq_update_affinity_hint(irq, NULL);
- free_irq(irq, gic);
- xa_erase(&gc->irq_contexts, i);
- kfree(gic);
+
+ mana_gd_put_gic(gc, false, i);
}
pci_free_irq_vectors(pdev);
--
2.43.0
^ permalink raw reply related
* [PATCH net-next v3 5/6] net: mana: Allocate interrupt context for each EQ when creating vPort
From: Long Li @ 2026-03-06 21:33 UTC (permalink / raw)
To: K . Y . Srinivasan, Haiyang Zhang, Wei Liu, Dexuan Cui,
David S . Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Shradha Gupta, Simon Horman, Konstantin Taranov,
Souradeep Chakrabarti, Erick Archer, linux-hyperv, netdev,
linux-kernel, linux-rdma
Cc: Long Li
In-Reply-To: <20260306213302.544681-1-longli@microsoft.com>
Use GIC functions to create a dedicated interrupt context or acquire a
shared interrupt context for each EQ when setting up a vPort.
Signed-off-by: Long Li <longli@microsoft.com>
---
drivers/net/ethernet/microsoft/mana/gdma_main.c | 2 +-
drivers/net/ethernet/microsoft/mana/mana_en.c | 17 ++++++++++++++++-
include/net/mana/gdma.h | 1 +
3 files changed, 18 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/microsoft/mana/gdma_main.c b/drivers/net/ethernet/microsoft/mana/gdma_main.c
index bdc9dc437fb7..81c0be96c94b 100644
--- a/drivers/net/ethernet/microsoft/mana/gdma_main.c
+++ b/drivers/net/ethernet/microsoft/mana/gdma_main.c
@@ -809,7 +809,6 @@ static void mana_gd_deregister_irq(struct gdma_queue *queue)
}
spin_unlock_irqrestore(&gic->lock, flags);
- queue->eq.msix_index = INVALID_PCI_MSIX_INDEX;
synchronize_rcu();
}
@@ -924,6 +923,7 @@ static int mana_gd_create_eq(struct gdma_dev *gd,
out:
dev_err(dev, "Failed to create EQ: %d\n", err);
mana_gd_destroy_eq(gc, false, queue);
+ queue->eq.msix_index = INVALID_PCI_MSIX_INDEX;
return err;
}
diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
index bfa0f354355d..8a60e7567951 100644
--- a/drivers/net/ethernet/microsoft/mana/mana_en.c
+++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
@@ -1598,6 +1598,7 @@ void mana_destroy_eq(struct mana_port_context *apc)
struct gdma_context *gc = ac->gdma_dev->gdma_context;
struct gdma_queue *eq;
int i;
+ unsigned int msi;
if (!apc->eqs)
return;
@@ -1610,7 +1611,9 @@ void mana_destroy_eq(struct mana_port_context *apc)
if (!eq)
continue;
+ msi = eq->eq.msix_index;
mana_gd_destroy_queue(gc, eq);
+ mana_gd_put_gic(gc, !gc->msi_sharing, msi);
}
kfree(apc->eqs);
@@ -1627,6 +1630,7 @@ static void mana_create_eq_debugfs(struct mana_port_context *apc, int i)
eq.mana_eq_debugfs = debugfs_create_dir(eqnum, apc->mana_eqs_debugfs);
debugfs_create_u32("head", 0400, eq.mana_eq_debugfs, &eq.eq->head);
debugfs_create_u32("tail", 0400, eq.mana_eq_debugfs, &eq.eq->tail);
+ debugfs_create_u32("irq", 0400, eq.mana_eq_debugfs, &eq.eq->eq.irq);
debugfs_create_file("eq_dump", 0400, eq.mana_eq_debugfs, eq.eq, &mana_dbg_q_fops);
}
@@ -1637,6 +1641,7 @@ int mana_create_eq(struct mana_port_context *apc)
struct gdma_queue_spec spec = {};
int err;
int i;
+ struct gdma_irq_context *gic;
WARN_ON(apc->eqs);
apc->eqs = kzalloc_objs(struct mana_eq, apc->num_queues);
@@ -1653,12 +1658,22 @@ int mana_create_eq(struct mana_port_context *apc)
apc->mana_eqs_debugfs = debugfs_create_dir("EQs", apc->mana_port_debugfs);
for (i = 0; i < apc->num_queues; i++) {
- spec.eq.msix_index = (i + 1) % gc->num_msix_usable;
+ if (gc->msi_sharing)
+ spec.eq.msix_index = (i + 1) % gc->num_msix_usable;
+
+ gic = mana_gd_get_gic(gc, !gc->msi_sharing, &spec.eq.msix_index);
+ if (!gic) {
+ err = -ENOMEM;
+ goto out;
+ }
+
err = mana_gd_create_mana_eq(gd, &spec, &apc->eqs[i].eq);
if (err) {
dev_err(gc->dev, "Failed to create EQ %d : %d\n", i, err);
+ mana_gd_put_gic(gc, !gc->msi_sharing, spec.eq.msix_index);
goto out;
}
+ apc->eqs[i].eq->eq.irq = gic->irq;
mana_create_eq_debugfs(apc, i);
}
diff --git a/include/net/mana/gdma.h b/include/net/mana/gdma.h
index 1a7f4abe7a8b..4e0278b00bbb 100644
--- a/include/net/mana/gdma.h
+++ b/include/net/mana/gdma.h
@@ -342,6 +342,7 @@ struct gdma_queue {
void *context;
unsigned int msix_index;
+ unsigned int irq;
u32 log2_throttle_limit;
} eq;
--
2.43.0
^ permalink raw reply related
* [PATCH net-next v3 6/6] RDMA/mana_ib: Allocate interrupt contexts on EQs
From: Long Li @ 2026-03-06 21:33 UTC (permalink / raw)
To: K . Y . Srinivasan, Haiyang Zhang, Wei Liu, Dexuan Cui,
David S . Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Shradha Gupta, Simon Horman, Konstantin Taranov,
Souradeep Chakrabarti, Erick Archer, linux-hyperv, netdev,
linux-kernel, linux-rdma
Cc: Long Li
In-Reply-To: <20260306213302.544681-1-longli@microsoft.com>
Use the GIC functions to allocate interrupt contexts for RDMA EQs. These
interrupt contexts may be shared with Ethernet EQs when MSI-X vectors
are limited.
The driver now supports allocating dedicated MSI-X for each EQ. Indicate
this capability through driver capability bits.
Signed-off-by: Long Li <longli@microsoft.com>
---
drivers/infiniband/hw/mana/main.c | 33 ++++++++++++++++++++++++++-----
include/net/mana/gdma.h | 7 +++++--
2 files changed, 33 insertions(+), 7 deletions(-)
diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c
index d51dd0ee85f4..0b74dd093b41 100644
--- a/drivers/infiniband/hw/mana/main.c
+++ b/drivers/infiniband/hw/mana/main.c
@@ -787,6 +787,7 @@ int mana_ib_create_eqs(struct mana_ib_dev *mdev)
{
struct gdma_context *gc = mdev_to_gc(mdev);
struct gdma_queue_spec spec = {};
+ struct gdma_irq_context *gic;
int err, i;
spec.type = GDMA_EQ;
@@ -797,9 +798,15 @@ int mana_ib_create_eqs(struct mana_ib_dev *mdev)
spec.eq.log2_throttle_limit = LOG2_EQ_THROTTLE;
spec.eq.msix_index = 0;
+ gic = mana_gd_get_gic(gc, false, &spec.eq.msix_index);
+ if (!gic)
+ return -ENOMEM;
+
err = mana_gd_create_mana_eq(mdev->gdma_dev, &spec, &mdev->fatal_err_eq);
- if (err)
+ if (err) {
+ mana_gd_put_gic(gc, false, 0);
return err;
+ }
mdev->eqs = kzalloc_objs(struct gdma_queue *,
mdev->ib_dev.num_comp_vectors);
@@ -810,31 +817,47 @@ int mana_ib_create_eqs(struct mana_ib_dev *mdev)
spec.eq.callback = NULL;
for (i = 0; i < mdev->ib_dev.num_comp_vectors; i++) {
spec.eq.msix_index = (i + 1) % gc->num_msix_usable;
+
+ gic = mana_gd_get_gic(gc, false, &spec.eq.msix_index);
+ if (!gic) {
+ err = -ENOMEM;
+ goto destroy_eqs;
+ }
+
err = mana_gd_create_mana_eq(mdev->gdma_dev, &spec, &mdev->eqs[i]);
- if (err)
+ if (err) {
+ mana_gd_put_gic(gc, false, spec.eq.msix_index);
goto destroy_eqs;
+ }
}
return 0;
destroy_eqs:
- while (i-- > 0)
+ while (i-- > 0) {
mana_gd_destroy_queue(gc, mdev->eqs[i]);
+ mana_gd_put_gic(gc, false, (i + 1) % gc->num_msix_usable);
+ }
kfree(mdev->eqs);
destroy_fatal_eq:
mana_gd_destroy_queue(gc, mdev->fatal_err_eq);
+ mana_gd_put_gic(gc, false, 0);
return err;
}
void mana_ib_destroy_eqs(struct mana_ib_dev *mdev)
{
struct gdma_context *gc = mdev_to_gc(mdev);
- int i;
+ int i, msi;
mana_gd_destroy_queue(gc, mdev->fatal_err_eq);
+ mana_gd_put_gic(gc, false, 0);
- for (i = 0; i < mdev->ib_dev.num_comp_vectors; i++)
+ for (i = 0; i < mdev->ib_dev.num_comp_vectors; i++) {
mana_gd_destroy_queue(gc, mdev->eqs[i]);
+ msi = (i + 1) % gc->num_msix_usable;
+ mana_gd_put_gic(gc, false, msi);
+ }
kfree(mdev->eqs);
}
diff --git a/include/net/mana/gdma.h b/include/net/mana/gdma.h
index 4e0278b00bbb..662e58f51e87 100644
--- a/include/net/mana/gdma.h
+++ b/include/net/mana/gdma.h
@@ -612,6 +612,7 @@ enum {
#define GDMA_DRV_CAP_FLAG_1_HWC_TIMEOUT_RECONFIG BIT(3)
#define GDMA_DRV_CAP_FLAG_1_GDMA_PAGES_4MB_1GB_2GB BIT(4)
#define GDMA_DRV_CAP_FLAG_1_VARIABLE_INDIRECTION_TABLE_SUPPORT BIT(5)
+#define GDMA_DRV_CAP_FLAG_1_HW_VPORT_LINK_AWARE BIT(6)
/* Driver can handle holes (zeros) in the device list */
#define GDMA_DRV_CAP_FLAG_1_DEV_LIST_HOLES_SUP BIT(11)
@@ -628,7 +629,8 @@ enum {
/* Driver detects stalled send queues and recovers them */
#define GDMA_DRV_CAP_FLAG_1_HANDLE_STALL_SQ_RECOVERY BIT(18)
-#define GDMA_DRV_CAP_FLAG_1_HW_VPORT_LINK_AWARE BIT(6)
+/* Driver supports separate EQ/MSIs for each vPort */
+#define GDMA_DRV_CAP_FLAG_1_EQ_MSI_UNSHARE_MULTI_VPORT BIT(19)
/* Driver supports linearizing the skb when num_sge exceeds hardware limit */
#define GDMA_DRV_CAP_FLAG_1_SKB_LINEARIZE BIT(20)
@@ -656,7 +658,8 @@ enum {
GDMA_DRV_CAP_FLAG_1_SKB_LINEARIZE | \
GDMA_DRV_CAP_FLAG_1_PROBE_RECOVERY | \
GDMA_DRV_CAP_FLAG_1_HANDLE_STALL_SQ_RECOVERY | \
- GDMA_DRV_CAP_FLAG_1_HWC_TIMEOUT_RECOVERY)
+ GDMA_DRV_CAP_FLAG_1_HWC_TIMEOUT_RECOVERY | \
+ GDMA_DRV_CAP_FLAG_1_EQ_MSI_UNSHARE_MULTI_VPORT)
#define GDMA_DRV_CAP_FLAGS2 0
--
2.43.0
^ permalink raw reply related
* [PATCH net-next,V3, 0/3] add ethtool COALESCE_RX_CQE_FRAMES/NSECS and use it in MANA driver
From: Haiyang Zhang @ 2026-03-06 23:19 UTC (permalink / raw)
To: linux-hyperv, netdev; +Cc: haiyangz, paulros
From: Haiyang Zhang <haiyangz@microsoft.com>
Add two parameters for drivers supporting Rx CQE Coalescing.
ETHTOOL_A_COALESCE_RX_CQE_FRAMES:
Maximum number of frames that can be coalesced into a CQE.
ETHTOOL_A_COALESCE_RX_CQE_NSECS:
Time out value in nanoseconds after the first packet arrival in a
coalesced CQE to be sent.
Also implement in MANA driver with the new parameter and
counters.
Haiyang Zhang (3):
net: ethtool: add ethtool COALESCE_RX_CQE_FRAMES/NSECS
net: mana: Add support for RX CQE Coalescing
net: mana: Add ethtool counters for RX CQEs in coalesced type
Documentation/netlink/specs/ethtool.yaml | 8 ++
Documentation/networking/ethtool-netlink.rst | 10 ++
drivers/net/ethernet/microsoft/mana/mana_en.c | 91 +++++++++++++------
.../ethernet/microsoft/mana/mana_ethtool.c | 75 ++++++++++++++-
include/linux/ethtool.h | 6 +-
include/net/mana/mana.h | 17 +++-
.../uapi/linux/ethtool_netlink_generated.h | 2 +
net/ethtool/coalesce.c | 14 ++-
8 files changed, 187 insertions(+), 36 deletions(-)
--
2.34.1
^ permalink raw reply
* [PATCH net-next,V3, 1/3] net: ethtool: add ethtool COALESCE_RX_CQE_FRAMES/NSECS
From: Haiyang Zhang @ 2026-03-06 23:19 UTC (permalink / raw)
To: linux-hyperv, netdev, Andrew Lunn, Jakub Kicinski,
David S. Miller, Eric Dumazet, Paolo Abeni, Simon Horman,
Donald Hunter, Jonathan Corbet, Shuah Khan,
Kory Maincent (Dent Project), Gal Pressman, Oleksij Rempel,
Vadim Fedorenko, linux-kernel, linux-doc
Cc: haiyangz, paulros
In-Reply-To: <20260306231936.549499-1-haiyangz@linux.microsoft.com>
From: Haiyang Zhang <haiyangz@microsoft.com>
Add two parameters for drivers supporting Rx CQE Coalescing.
ETHTOOL_A_COALESCE_RX_CQE_FRAMES:
Maximum number of frames that can be coalesced into a CQE.
ETHTOOL_A_COALESCE_RX_CQE_NSECS:
Time out value in nanoseconds after the first packet arrival in a
coalesced CQE to be sent.
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
---
Documentation/netlink/specs/ethtool.yaml | 8 ++++++++
Documentation/networking/ethtool-netlink.rst | 10 ++++++++++
include/linux/ethtool.h | 6 +++++-
include/uapi/linux/ethtool_netlink_generated.h | 2 ++
net/ethtool/coalesce.c | 14 +++++++++++++-
5 files changed, 38 insertions(+), 2 deletions(-)
diff --git a/Documentation/netlink/specs/ethtool.yaml b/Documentation/netlink/specs/ethtool.yaml
index 4707063af3b4..d254e26c014c 100644
--- a/Documentation/netlink/specs/ethtool.yaml
+++ b/Documentation/netlink/specs/ethtool.yaml
@@ -861,6 +861,12 @@ attribute-sets:
name: tx-profile
type: nest
nested-attributes: profile
+ -
+ name: rx-cqe-frames
+ type: u32
+ -
+ name: rx-cqe-nsecs
+ type: u32
-
name: pause-stat
@@ -2257,6 +2263,8 @@ operations:
- tx-aggr-time-usecs
- rx-profile
- tx-profile
+ - rx-cqe-frames
+ - rx-cqe-nsecs
dump: *coalesce-get-op
-
name: coalesce-set
diff --git a/Documentation/networking/ethtool-netlink.rst b/Documentation/networking/ethtool-netlink.rst
index 32179168eb73..a9fbb16891fa 100644
--- a/Documentation/networking/ethtool-netlink.rst
+++ b/Documentation/networking/ethtool-netlink.rst
@@ -1076,6 +1076,8 @@ Kernel response contents:
``ETHTOOL_A_COALESCE_TX_AGGR_TIME_USECS`` u32 time (us), aggr, Tx
``ETHTOOL_A_COALESCE_RX_PROFILE`` nested profile of DIM, Rx
``ETHTOOL_A_COALESCE_TX_PROFILE`` nested profile of DIM, Tx
+ ``ETHTOOL_A_COALESCE_RX_CQE_FRAMES`` u32 max packets, Rx CQE
+ ``ETHTOOL_A_COALESCE_RX_CQE_NSECS`` u32 delay (ns), Rx CQE
=========================================== ====== =======================
Attributes are only included in reply if their value is not zero or the
@@ -1109,6 +1111,12 @@ well with frequent small-sized URBs transmissions.
to DIM parameters, see `Generic Network Dynamic Interrupt Moderation (Net DIM)
<https://www.kernel.org/doc/Documentation/networking/net_dim.rst>`_.
+Rx CQE coalescing allows multiple received packets to be coalesced into a single
+Completion Queue Entry (CQE). ``ETHTOOL_A_COALESCE_RX_CQE_FRAMES`` describes the
+maximum number of frames that can be coalesced into a CQE.
+``ETHTOOL_A_COALESCE_RX_CQE_NSECS`` describes max time in nanoseconds after the
+first packet arrival in a coalesced CQE to be sent.
+
COALESCE_SET
============
@@ -1147,6 +1155,8 @@ Request contents:
``ETHTOOL_A_COALESCE_TX_AGGR_TIME_USECS`` u32 time (us), aggr, Tx
``ETHTOOL_A_COALESCE_RX_PROFILE`` nested profile of DIM, Rx
``ETHTOOL_A_COALESCE_TX_PROFILE`` nested profile of DIM, Tx
+ ``ETHTOOL_A_COALESCE_RX_CQE_FRAMES`` u32 max packets, Rx CQE
+ ``ETHTOOL_A_COALESCE_RX_CQE_NSECS`` u32 delay (ns), Rx CQE
=========================================== ====== =======================
Request is rejected if it attributes declared as unsupported by driver (i.e.
diff --git a/include/linux/ethtool.h b/include/linux/ethtool.h
index 83c375840835..656d465bcd06 100644
--- a/include/linux/ethtool.h
+++ b/include/linux/ethtool.h
@@ -332,6 +332,8 @@ struct kernel_ethtool_coalesce {
u32 tx_aggr_max_bytes;
u32 tx_aggr_max_frames;
u32 tx_aggr_time_usecs;
+ u32 rx_cqe_frames;
+ u32 rx_cqe_nsecs;
};
/**
@@ -380,7 +382,9 @@ bool ethtool_convert_link_mode_to_legacy_u32(u32 *legacy_u32,
#define ETHTOOL_COALESCE_TX_AGGR_TIME_USECS BIT(26)
#define ETHTOOL_COALESCE_RX_PROFILE BIT(27)
#define ETHTOOL_COALESCE_TX_PROFILE BIT(28)
-#define ETHTOOL_COALESCE_ALL_PARAMS GENMASK(28, 0)
+#define ETHTOOL_COALESCE_RX_CQE_FRAMES BIT(29)
+#define ETHTOOL_COALESCE_RX_CQE_NSECS BIT(30)
+#define ETHTOOL_COALESCE_ALL_PARAMS GENMASK(30, 0)
#define ETHTOOL_COALESCE_USECS \
(ETHTOOL_COALESCE_RX_USECS | ETHTOOL_COALESCE_TX_USECS)
diff --git a/include/uapi/linux/ethtool_netlink_generated.h b/include/uapi/linux/ethtool_netlink_generated.h
index 114b83017297..8134baf7860f 100644
--- a/include/uapi/linux/ethtool_netlink_generated.h
+++ b/include/uapi/linux/ethtool_netlink_generated.h
@@ -371,6 +371,8 @@ enum {
ETHTOOL_A_COALESCE_TX_AGGR_TIME_USECS,
ETHTOOL_A_COALESCE_RX_PROFILE,
ETHTOOL_A_COALESCE_TX_PROFILE,
+ ETHTOOL_A_COALESCE_RX_CQE_FRAMES,
+ ETHTOOL_A_COALESCE_RX_CQE_NSECS,
__ETHTOOL_A_COALESCE_CNT,
ETHTOOL_A_COALESCE_MAX = (__ETHTOOL_A_COALESCE_CNT - 1)
diff --git a/net/ethtool/coalesce.c b/net/ethtool/coalesce.c
index 3e18ca1ccc5e..349bb02c517a 100644
--- a/net/ethtool/coalesce.c
+++ b/net/ethtool/coalesce.c
@@ -118,6 +118,8 @@ static int coalesce_reply_size(const struct ethnl_req_info *req_base,
nla_total_size(sizeof(u32)) + /* _TX_AGGR_MAX_BYTES */
nla_total_size(sizeof(u32)) + /* _TX_AGGR_MAX_FRAMES */
nla_total_size(sizeof(u32)) + /* _TX_AGGR_TIME_USECS */
+ nla_total_size(sizeof(u32)) + /* _RX_CQE_FRAMES */
+ nla_total_size(sizeof(u32)) + /* _RX_CQE_NSECS */
total_modersz * 2; /* _{R,T}X_PROFILE */
}
@@ -269,7 +271,11 @@ static int coalesce_fill_reply(struct sk_buff *skb,
coalesce_put_u32(skb, ETHTOOL_A_COALESCE_TX_AGGR_MAX_FRAMES,
kcoal->tx_aggr_max_frames, supported) ||
coalesce_put_u32(skb, ETHTOOL_A_COALESCE_TX_AGGR_TIME_USECS,
- kcoal->tx_aggr_time_usecs, supported))
+ kcoal->tx_aggr_time_usecs, supported) ||
+ coalesce_put_u32(skb, ETHTOOL_A_COALESCE_RX_CQE_FRAMES,
+ kcoal->rx_cqe_frames, supported) ||
+ coalesce_put_u32(skb, ETHTOOL_A_COALESCE_RX_CQE_NSECS,
+ kcoal->rx_cqe_nsecs, supported))
return -EMSGSIZE;
if (!req_base->dev || !req_base->dev->irq_moder)
@@ -338,6 +344,8 @@ const struct nla_policy ethnl_coalesce_set_policy[] = {
[ETHTOOL_A_COALESCE_TX_AGGR_MAX_BYTES] = { .type = NLA_U32 },
[ETHTOOL_A_COALESCE_TX_AGGR_MAX_FRAMES] = { .type = NLA_U32 },
[ETHTOOL_A_COALESCE_TX_AGGR_TIME_USECS] = { .type = NLA_U32 },
+ [ETHTOOL_A_COALESCE_RX_CQE_FRAMES] = { .type = NLA_U32 },
+ [ETHTOOL_A_COALESCE_RX_CQE_NSECS] = { .type = NLA_U32 },
[ETHTOOL_A_COALESCE_RX_PROFILE] =
NLA_POLICY_NESTED(coalesce_profile_policy),
[ETHTOOL_A_COALESCE_TX_PROFILE] =
@@ -570,6 +578,10 @@ __ethnl_set_coalesce(struct ethnl_req_info *req_info, struct genl_info *info,
tb[ETHTOOL_A_COALESCE_TX_AGGR_MAX_FRAMES], &mod);
ethnl_update_u32(&kernel_coalesce.tx_aggr_time_usecs,
tb[ETHTOOL_A_COALESCE_TX_AGGR_TIME_USECS], &mod);
+ ethnl_update_u32(&kernel_coalesce.rx_cqe_frames,
+ tb[ETHTOOL_A_COALESCE_RX_CQE_FRAMES], &mod);
+ ethnl_update_u32(&kernel_coalesce.rx_cqe_nsecs,
+ tb[ETHTOOL_A_COALESCE_RX_CQE_NSECS], &mod);
if (dev->irq_moder && dev->irq_moder->profile_flags & DIM_PROFILE_RX) {
ret = ethnl_update_profile(dev, &dev->irq_moder->rx_profile,
--
2.34.1
^ permalink raw reply related
* [PATCH net-next,V3, 2/3] net: mana: Add support for RX CQE Coalescing
From: Haiyang Zhang @ 2026-03-06 23:19 UTC (permalink / raw)
To: linux-hyperv, netdev, K. Y. Srinivasan, Haiyang Zhang, Wei Liu,
Dexuan Cui, Long Li, Andrew Lunn, David S. Miller, Eric Dumazet,
Jakub Kicinski, Paolo Abeni, Konstantin Taranov, Simon Horman,
Erni Sri Satya Vennela, Shradha Gupta, Dipayaan Roy,
Shiraz Saleem, Kees Cook, Subbaraya Sundeep, Breno Leitao,
Aditya Garg, linux-kernel, linux-rdma
Cc: paulros
In-Reply-To: <20260306231936.549499-1-haiyangz@linux.microsoft.com>
From: Haiyang Zhang <haiyangz@microsoft.com>
Our NIC can have up to 4 RX packets on 1 CQE. To support this feature,
check and process the type CQE_RX_COALESCED_4. The default setting is
disabled, to avoid possible regression on latency.
And, add ethtool handler to switch this feature. To turn it on, run:
ethtool -C <nic> rx-cqe-frames 4
To turn it off:
ethtool -C <nic> rx-cqe-frames 1
The rx-cqe-nsec is the time out value in nanoseconds after the first
packet arrival in a coalesced CQE to be sent. It's read-only for this
NIC.
Reviewed-by: Long Li <longli@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
---
drivers/net/ethernet/microsoft/mana/mana_en.c | 72 ++++++++++++-------
.../ethernet/microsoft/mana/mana_ethtool.c | 60 +++++++++++++++-
include/net/mana/mana.h | 8 ++-
3 files changed, 111 insertions(+), 29 deletions(-)
diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
index ea71de39f996..c06fec50e51f 100644
--- a/drivers/net/ethernet/microsoft/mana/mana_en.c
+++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
@@ -1365,6 +1365,7 @@ static int mana_cfg_vport_steering(struct mana_port_context *apc,
sizeof(resp));
req->hdr.req.msg_version = GDMA_MESSAGE_V2;
+ req->hdr.resp.msg_version = GDMA_MESSAGE_V2;
req->vport = apc->port_handle;
req->num_indir_entries = apc->indir_table_sz;
@@ -1376,7 +1377,9 @@ static int mana_cfg_vport_steering(struct mana_port_context *apc,
req->update_hashkey = update_key;
req->update_indir_tab = update_tab;
req->default_rxobj = apc->default_rxobj;
- req->cqe_coalescing_enable = 0;
+
+ if (rx != TRI_STATE_FALSE)
+ req->cqe_coalescing_enable = apc->cqe_coalescing_enable;
if (update_key)
memcpy(&req->hashkey, apc->hashkey, MANA_HASH_KEY_SIZE);
@@ -1407,6 +1410,10 @@ static int mana_cfg_vport_steering(struct mana_port_context *apc,
err = -EPROTO;
}
+ if (resp.hdr.response.msg_version >= GDMA_MESSAGE_V2)
+ apc->cqe_coalescing_timeout_ns =
+ resp.cqe_coalescing_timeout_ns;
+
netdev_info(ndev, "Configured steering vPort %llu entries %u\n",
apc->port_handle, apc->indir_table_sz);
out:
@@ -1915,11 +1922,12 @@ static struct sk_buff *mana_build_skb(struct mana_rxq *rxq, void *buf_va,
}
static void mana_rx_skb(void *buf_va, bool from_pool,
- struct mana_rxcomp_oob *cqe, struct mana_rxq *rxq)
+ struct mana_rxcomp_oob *cqe, struct mana_rxq *rxq,
+ int i)
{
struct mana_stats_rx *rx_stats = &rxq->stats;
struct net_device *ndev = rxq->ndev;
- uint pkt_len = cqe->ppi[0].pkt_len;
+ uint pkt_len = cqe->ppi[i].pkt_len;
u16 rxq_idx = rxq->rxq_idx;
struct napi_struct *napi;
struct xdp_buff xdp = {};
@@ -1963,7 +1971,7 @@ static void mana_rx_skb(void *buf_va, bool from_pool,
}
if (cqe->rx_hashtype != 0 && (ndev->features & NETIF_F_RXHASH)) {
- hash_value = cqe->ppi[0].pkt_hash;
+ hash_value = cqe->ppi[i].pkt_hash;
if (cqe->rx_hashtype & MANA_HASH_L4)
skb_set_hash(skb, hash_value, PKT_HASH_TYPE_L4);
@@ -2098,9 +2106,11 @@ static void mana_process_rx_cqe(struct mana_rxq *rxq, struct mana_cq *cq,
struct mana_recv_buf_oob *rxbuf_oob;
struct mana_port_context *apc;
struct device *dev = gc->dev;
+ bool coalesced = false;
void *old_buf = NULL;
u32 curr, pktlen;
bool old_fp;
+ int i;
apc = netdev_priv(ndev);
@@ -2112,13 +2122,16 @@ static void mana_process_rx_cqe(struct mana_rxq *rxq, struct mana_cq *cq,
++ndev->stats.rx_dropped;
rxbuf_oob = &rxq->rx_oobs[rxq->buf_index];
netdev_warn_once(ndev, "Dropped a truncated packet\n");
- goto drop;
- case CQE_RX_COALESCED_4:
- netdev_err(ndev, "RX coalescing is unsupported\n");
- apc->eth_stats.rx_coalesced_err++;
+ mana_move_wq_tail(rxq->gdma_rq,
+ rxbuf_oob->wqe_inf.wqe_size_in_bu);
+ mana_post_pkt_rxq(rxq);
return;
+ case CQE_RX_COALESCED_4:
+ coalesced = true;
+ break;
+
case CQE_RX_OBJECT_FENCE:
complete(&rxq->fence_event);
return;
@@ -2130,30 +2143,36 @@ static void mana_process_rx_cqe(struct mana_rxq *rxq, struct mana_cq *cq,
return;
}
- pktlen = oob->ppi[0].pkt_len;
+ for (i = 0; i < MANA_RXCOMP_OOB_NUM_PPI; i++) {
+ pktlen = oob->ppi[i].pkt_len;
+ if (pktlen == 0) {
+ if (i == 0)
+ netdev_err_once(
+ ndev,
+ "RX pkt len=0, rq=%u, cq=%u, rxobj=0x%llx\n",
+ rxq->gdma_id, cq->gdma_id, rxq->rxobj);
+ break;
+ }
- if (pktlen == 0) {
- /* data packets should never have packetlength of zero */
- netdev_err(ndev, "RX pkt len=0, rq=%u, cq=%u, rxobj=0x%llx\n",
- rxq->gdma_id, cq->gdma_id, rxq->rxobj);
- return;
- }
+ curr = rxq->buf_index;
+ rxbuf_oob = &rxq->rx_oobs[curr];
+ WARN_ON_ONCE(rxbuf_oob->wqe_inf.wqe_size_in_bu != 1);
- curr = rxq->buf_index;
- rxbuf_oob = &rxq->rx_oobs[curr];
- WARN_ON_ONCE(rxbuf_oob->wqe_inf.wqe_size_in_bu != 1);
+ mana_refill_rx_oob(dev, rxq, rxbuf_oob, &old_buf, &old_fp);
- mana_refill_rx_oob(dev, rxq, rxbuf_oob, &old_buf, &old_fp);
+ /* Unsuccessful refill will have old_buf == NULL.
+ * In this case, mana_rx_skb() will drop the packet.
+ */
+ mana_rx_skb(old_buf, old_fp, oob, rxq, i);
- /* Unsuccessful refill will have old_buf == NULL.
- * In this case, mana_rx_skb() will drop the packet.
- */
- mana_rx_skb(old_buf, old_fp, oob, rxq);
+ mana_move_wq_tail(rxq->gdma_rq,
+ rxbuf_oob->wqe_inf.wqe_size_in_bu);
-drop:
- mana_move_wq_tail(rxq->gdma_rq, rxbuf_oob->wqe_inf.wqe_size_in_bu);
+ mana_post_pkt_rxq(rxq);
- mana_post_pkt_rxq(rxq);
+ if (!coalesced)
+ break;
+ }
}
static void mana_poll_rx_cq(struct mana_cq *cq)
@@ -3332,6 +3351,7 @@ static int mana_probe_port(struct mana_context *ac, int port_idx,
apc->port_handle = INVALID_MANA_HANDLE;
apc->pf_filter_handle = INVALID_MANA_HANDLE;
apc->port_idx = port_idx;
+ apc->cqe_coalescing_enable = 0;
mutex_init(&apc->vport_mutex);
apc->vport_use_count = 0;
diff --git a/drivers/net/ethernet/microsoft/mana/mana_ethtool.c b/drivers/net/ethernet/microsoft/mana/mana_ethtool.c
index f2d220b371b5..4b234b16e57a 100644
--- a/drivers/net/ethernet/microsoft/mana/mana_ethtool.c
+++ b/drivers/net/ethernet/microsoft/mana/mana_ethtool.c
@@ -20,8 +20,6 @@ static const struct mana_stats_desc mana_eth_stats[] = {
tx_cqe_unknown_type)},
{"tx_linear_pkt_cnt", offsetof(struct mana_ethtool_stats,
tx_linear_pkt_cnt)},
- {"rx_coalesced_err", offsetof(struct mana_ethtool_stats,
- rx_coalesced_err)},
{"rx_cqe_unknown_type", offsetof(struct mana_ethtool_stats,
rx_cqe_unknown_type)},
};
@@ -390,6 +388,61 @@ static void mana_get_channels(struct net_device *ndev,
channel->combined_count = apc->num_queues;
}
+#define MANA_RX_CQE_NSEC_DEF 2048
+static int mana_get_coalesce(struct net_device *ndev,
+ struct ethtool_coalesce *ec,
+ struct kernel_ethtool_coalesce *kernel_coal,
+ struct netlink_ext_ack *extack)
+{
+ struct mana_port_context *apc = netdev_priv(ndev);
+
+ kernel_coal->rx_cqe_frames =
+ apc->cqe_coalescing_enable ? MANA_RXCOMP_OOB_NUM_PPI : 1;
+
+ kernel_coal->rx_cqe_nsecs = apc->cqe_coalescing_timeout_ns;
+
+ /* Return the default timeout value for old FW not providing
+ * this value.
+ */
+ if (apc->port_is_up && apc->cqe_coalescing_enable &&
+ !kernel_coal->rx_cqe_nsecs)
+ kernel_coal->rx_cqe_nsecs = MANA_RX_CQE_NSEC_DEF;
+
+ return 0;
+}
+
+static int mana_set_coalesce(struct net_device *ndev,
+ struct ethtool_coalesce *ec,
+ struct kernel_ethtool_coalesce *kernel_coal,
+ struct netlink_ext_ack *extack)
+{
+ struct mana_port_context *apc = netdev_priv(ndev);
+ u8 saved_cqe_coalescing_enable;
+ int err;
+
+ if (kernel_coal->rx_cqe_frames != 1 &&
+ kernel_coal->rx_cqe_frames != MANA_RXCOMP_OOB_NUM_PPI) {
+ NL_SET_ERR_MSG_FMT(extack,
+ "rx-frames must be 1 or %u, got %u",
+ MANA_RXCOMP_OOB_NUM_PPI,
+ kernel_coal->rx_cqe_frames);
+ return -EINVAL;
+ }
+
+ saved_cqe_coalescing_enable = apc->cqe_coalescing_enable;
+ apc->cqe_coalescing_enable =
+ kernel_coal->rx_cqe_frames == MANA_RXCOMP_OOB_NUM_PPI;
+
+ if (!apc->port_is_up)
+ return 0;
+
+ err = mana_config_rss(apc, TRI_STATE_TRUE, false, false);
+ if (err)
+ apc->cqe_coalescing_enable = saved_cqe_coalescing_enable;
+
+ return err;
+}
+
static int mana_set_channels(struct net_device *ndev,
struct ethtool_channels *channels)
{
@@ -510,6 +563,7 @@ static int mana_get_link_ksettings(struct net_device *ndev,
}
const struct ethtool_ops mana_ethtool_ops = {
+ .supported_coalesce_params = ETHTOOL_COALESCE_RX_CQE_FRAMES,
.get_ethtool_stats = mana_get_ethtool_stats,
.get_sset_count = mana_get_sset_count,
.get_strings = mana_get_strings,
@@ -520,6 +574,8 @@ const struct ethtool_ops mana_ethtool_ops = {
.set_rxfh = mana_set_rxfh,
.get_channels = mana_get_channels,
.set_channels = mana_set_channels,
+ .get_coalesce = mana_get_coalesce,
+ .set_coalesce = mana_set_coalesce,
.get_ringparam = mana_get_ringparam,
.set_ringparam = mana_set_ringparam,
.get_link_ksettings = mana_get_link_ksettings,
diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h
index a078af283bdd..a7f89e7ddc56 100644
--- a/include/net/mana/mana.h
+++ b/include/net/mana/mana.h
@@ -378,7 +378,6 @@ struct mana_ethtool_stats {
u64 tx_cqe_err;
u64 tx_cqe_unknown_type;
u64 tx_linear_pkt_cnt;
- u64 rx_coalesced_err;
u64 rx_cqe_unknown_type;
};
@@ -557,6 +556,9 @@ struct mana_port_context {
bool port_is_up;
bool port_st_save; /* Saved port state */
+ u8 cqe_coalescing_enable;
+ u32 cqe_coalescing_timeout_ns;
+
struct mana_ethtool_stats eth_stats;
struct mana_ethtool_phy_stats phy_stats;
@@ -902,6 +904,10 @@ struct mana_cfg_rx_steer_req_v2 {
struct mana_cfg_rx_steer_resp {
struct gdma_resp_hdr hdr;
+
+ /* V2 */
+ u32 cqe_coalescing_timeout_ns;
+ u32 reserved1;
}; /* HW DATA */
/* Register HW vPort */
--
2.34.1
^ permalink raw reply related
* [PATCH net-next,V3, 3/3] net: mana: Add ethtool counters for RX CQEs in coalesced type
From: Haiyang Zhang @ 2026-03-06 23:19 UTC (permalink / raw)
To: linux-hyperv, netdev, K. Y. Srinivasan, Haiyang Zhang, Wei Liu,
Dexuan Cui, Long Li, Andrew Lunn, David S. Miller, Eric Dumazet,
Jakub Kicinski, Paolo Abeni, Konstantin Taranov, Simon Horman,
Erni Sri Satya Vennela, Dipayaan Roy, Shradha Gupta,
Shiraz Saleem, Kees Cook, Subbaraya Sundeep, Aditya Garg,
Breno Leitao, linux-kernel, linux-rdma
Cc: paulros
In-Reply-To: <20260306231936.549499-1-haiyangz@linux.microsoft.com>
From: Haiyang Zhang <haiyangz@microsoft.com>
For RX CQEs with type CQE_RX_COALESCED_4, to measure the coalescing
efficiency, add counters to count how many contains 2, 3, 4 packets
respectively.
Also, add a counter for the error case of first packet with length == 0.
Reviewed-by: Long Li <longli@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
---
drivers/net/ethernet/microsoft/mana/mana_en.c | 21 ++++++++++++++++++-
.../ethernet/microsoft/mana/mana_ethtool.c | 15 +++++++++++--
include/net/mana/mana.h | 9 +++++---
3 files changed, 39 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
index c06fec50e51f..11ea2b17502d 100644
--- a/drivers/net/ethernet/microsoft/mana/mana_en.c
+++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
@@ -2146,11 +2146,23 @@ static void mana_process_rx_cqe(struct mana_rxq *rxq, struct mana_cq *cq,
for (i = 0; i < MANA_RXCOMP_OOB_NUM_PPI; i++) {
pktlen = oob->ppi[i].pkt_len;
if (pktlen == 0) {
- if (i == 0)
+ /* Collect coalesced CQE count based on packets processed.
+ * Coalesced CQEs have at least 2 packets, so index is i - 2.
+ */
+ if (i > 1) {
+ u64_stats_update_begin(&rxq->stats.syncp);
+ rxq->stats.coalesced_cqe[i - 2]++;
+ u64_stats_update_end(&rxq->stats.syncp);
+ } else if (i == 0) {
+ /* Error case stat */
+ u64_stats_update_begin(&rxq->stats.syncp);
+ rxq->stats.pkt_len0_err++;
+ u64_stats_update_end(&rxq->stats.syncp);
netdev_err_once(
ndev,
"RX pkt len=0, rq=%u, cq=%u, rxobj=0x%llx\n",
rxq->gdma_id, cq->gdma_id, rxq->rxobj);
+ }
break;
}
@@ -2173,6 +2185,13 @@ static void mana_process_rx_cqe(struct mana_rxq *rxq, struct mana_cq *cq,
if (!coalesced)
break;
}
+
+ /* Coalesced CQE with all 4 packets */
+ if (coalesced && i == MANA_RXCOMP_OOB_NUM_PPI) {
+ u64_stats_update_begin(&rxq->stats.syncp);
+ rxq->stats.coalesced_cqe[MANA_RXCOMP_OOB_NUM_PPI - 2]++;
+ u64_stats_update_end(&rxq->stats.syncp);
+ }
}
static void mana_poll_rx_cq(struct mana_cq *cq)
diff --git a/drivers/net/ethernet/microsoft/mana/mana_ethtool.c b/drivers/net/ethernet/microsoft/mana/mana_ethtool.c
index 4b234b16e57a..6a4b42fe0944 100644
--- a/drivers/net/ethernet/microsoft/mana/mana_ethtool.c
+++ b/drivers/net/ethernet/microsoft/mana/mana_ethtool.c
@@ -149,7 +149,7 @@ static void mana_get_strings(struct net_device *ndev, u32 stringset, u8 *data)
{
struct mana_port_context *apc = netdev_priv(ndev);
unsigned int num_queues = apc->num_queues;
- int i;
+ int i, j;
if (stringset != ETH_SS_STATS)
return;
@@ -168,6 +168,9 @@ static void mana_get_strings(struct net_device *ndev, u32 stringset, u8 *data)
ethtool_sprintf(&data, "rx_%d_xdp_drop", i);
ethtool_sprintf(&data, "rx_%d_xdp_tx", i);
ethtool_sprintf(&data, "rx_%d_xdp_redirect", i);
+ ethtool_sprintf(&data, "rx_%d_pkt_len0_err", i);
+ for (j = 0; j < MANA_RXCOMP_OOB_NUM_PPI - 1; j++)
+ ethtool_sprintf(&data, "rx_%d_coalesced_cqe_%d", i, j + 2);
}
for (i = 0; i < num_queues; i++) {
@@ -201,6 +204,8 @@ static void mana_get_ethtool_stats(struct net_device *ndev,
u64 xdp_xmit;
u64 xdp_drop;
u64 xdp_tx;
+ u64 pkt_len0_err;
+ u64 coalesced_cqe[MANA_RXCOMP_OOB_NUM_PPI - 1];
u64 tso_packets;
u64 tso_bytes;
u64 tso_inner_packets;
@@ -209,7 +214,7 @@ static void mana_get_ethtool_stats(struct net_device *ndev,
u64 short_pkt_fmt;
u64 csum_partial;
u64 mana_map_err;
- int q, i = 0;
+ int q, i = 0, j;
if (!apc->port_is_up)
return;
@@ -239,6 +244,9 @@ static void mana_get_ethtool_stats(struct net_device *ndev,
xdp_drop = rx_stats->xdp_drop;
xdp_tx = rx_stats->xdp_tx;
xdp_redirect = rx_stats->xdp_redirect;
+ pkt_len0_err = rx_stats->pkt_len0_err;
+ for (j = 0; j < MANA_RXCOMP_OOB_NUM_PPI - 1; j++)
+ coalesced_cqe[j] = rx_stats->coalesced_cqe[j];
} while (u64_stats_fetch_retry(&rx_stats->syncp, start));
data[i++] = packets;
@@ -246,6 +254,9 @@ static void mana_get_ethtool_stats(struct net_device *ndev,
data[i++] = xdp_drop;
data[i++] = xdp_tx;
data[i++] = xdp_redirect;
+ data[i++] = pkt_len0_err;
+ for (j = 0; j < MANA_RXCOMP_OOB_NUM_PPI - 1; j++)
+ data[i++] = coalesced_cqe[j];
}
for (q = 0; q < num_queues; q++) {
diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h
index a7f89e7ddc56..3336688fed5e 100644
--- a/include/net/mana/mana.h
+++ b/include/net/mana/mana.h
@@ -61,8 +61,11 @@ enum TRI_STATE {
#define MAX_PORTS_IN_MANA_DEV 256
+/* Maximum number of packets per coalesced CQE */
+#define MANA_RXCOMP_OOB_NUM_PPI 4
+
/* Update this count whenever the respective structures are changed */
-#define MANA_STATS_RX_COUNT 5
+#define MANA_STATS_RX_COUNT (6 + MANA_RXCOMP_OOB_NUM_PPI - 1)
#define MANA_STATS_TX_COUNT 11
#define MANA_RX_FRAG_ALIGNMENT 64
@@ -73,6 +76,8 @@ struct mana_stats_rx {
u64 xdp_drop;
u64 xdp_tx;
u64 xdp_redirect;
+ u64 pkt_len0_err;
+ u64 coalesced_cqe[MANA_RXCOMP_OOB_NUM_PPI - 1];
struct u64_stats_sync syncp;
};
@@ -227,8 +232,6 @@ struct mana_rxcomp_perpkt_info {
u32 pkt_hash;
}; /* HW DATA */
-#define MANA_RXCOMP_OOB_NUM_PPI 4
-
/* Receive completion OOB */
struct mana_rxcomp_oob {
struct mana_cqe_header cqe_hdr;
--
2.34.1
^ permalink raw reply related
* [PATCH 0/8] RDMA/mana_ib: Handle service reset for RDMA resources
From: Long Li @ 2026-03-07 1:44 UTC (permalink / raw)
To: K . Y . Srinivasan, Haiyang Zhang, Wei Liu, Dexuan Cui,
David S . Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Shradha Gupta, Simon Horman, Konstantin Taranov,
Souradeep Chakrabarti, Erick Archer, linux-hyperv, netdev,
linux-kernel, linux-rdma
Cc: Long Li
When the MANA hardware undergoes a service reset, the ETH auxiliary device
(mana.eth) used by DPDK persists across the reset cycle — it is not removed
and re-added like RC/UD/GSI QPs. This means userspace RDMA consumers such
as DPDK have no way of knowing that firmware handles for their PD, CQ, WQ,
QP and MR resources have become stale.
This series adds per-ucontext resource tracking and a reset notification
mechanism so that:
1. The RDMA driver is informed of service reset events via direct callbacks
from the ETH driver (reset_notify / resume_notify).
2. On reset, all tracked firmware handles are invalidated (set to
INVALID_MANA_HANDLE), user doorbell mappings are revoked via
rdma_user_mmap_disassociate(), and IB_EVENT_PORT_ERR is dispatched to
each affected ucontext so userspace can detect the reset.
3. Destroy callbacks check for INVALID_MANA_HANDLE and skip firmware
commands for resources already invalidated by the reset path,
preventing stale handles from being sent to firmware.
4. A reset_rwsem serializes handle invalidation against resource creation
to avoid races between the reset path and new resource allocation.
Patches 1-6 introduce per-ucontext tracking lists for each resource type.
Patch 7 implements the reset/resume notification mechanism with rwsem
serialization, mmap revocation, and IB event dispatch.
Patch 8 adds INVALID_MANA_HANDLE checks in destroy callbacks.
Tested with DPDK testpmd on Azure VM (linux-next-20260306) — confirmed
IB_EVENT_PORT_ERR (type=10) and IB_EVENT_PORT_ACTIVE (type=9) are delivered
to userspace during service reset, and testpmd tears down cleanly afterwards.
Long Li (8):
RDMA/mana_ib: Track ucontext per device
RDMA/mana_ib: Track PD per ucontext
RDMA/mana_ib: Track CQ per ucontext
RDMA/mana_ib: Track WQ per ucontext
RDMA/mana_ib: Track QP per ucontext
RDMA/mana_ib: Track MR per ucontext
RDMA/mana_ib: Notify service reset events to RDMA devices
RDMA/mana_ib: Skip firmware commands for invalidated handles
drivers/infiniband/hw/mana/cq.c | 44 +++++--
drivers/infiniband/hw/mana/device.c | 105 ++++++++++++++++++
drivers/infiniband/hw/mana/main.c | 56 +++++++++-
drivers/infiniband/hw/mana/mana_ib.h | 19 ++++
drivers/infiniband/hw/mana/mr.c | 33 +++++-
drivers/infiniband/hw/mana/qp.c | 61 +++++++---
drivers/infiniband/hw/mana/wq.c | 24 ++++
drivers/net/ethernet/microsoft/mana/mana_en.c | 14 ++-
include/net/mana/gdma.h | 6 +
9 files changed, 331 insertions(+), 31 deletions(-)
--
2.43.0
^ permalink raw reply
* [PATCH rdma-next 0/8] RDMA/mana_ib: Handle service reset for RDMA resources
From: Long Li @ 2026-03-07 1:47 UTC (permalink / raw)
To: Long Li, Konstantin Taranov, Jakub Kicinski, David S . Miller,
Paolo Abeni, Eric Dumazet, Andrew Lunn, Jason Gunthorpe,
Leon Romanovsky, Haiyang Zhang, K . Y . Srinivasan, Wei Liu,
Dexuan Cui
Cc: Simon Horman, netdev, linux-rdma, linux-hyperv, linux-kernel
When the MANA hardware undergoes a service reset, the ETH auxiliary device
(mana.eth) used by DPDK persists across the reset cycle — it is not removed
and re-added like RC/UD/GSI QPs. This means userspace RDMA consumers such
as DPDK have no way of knowing that firmware handles for their PD, CQ, WQ,
QP and MR resources have become stale.
This series adds per-ucontext resource tracking and a reset notification
mechanism so that:
1. The RDMA driver is informed of service reset events via direct callbacks
from the ETH driver (reset_notify / resume_notify).
2. On reset, all tracked firmware handles are invalidated (set to
INVALID_MANA_HANDLE), user doorbell mappings are revoked via
rdma_user_mmap_disassociate(), and IB_EVENT_PORT_ERR is dispatched to
each affected ucontext so userspace can detect the reset.
3. Destroy callbacks check for INVALID_MANA_HANDLE and skip firmware
commands for resources already invalidated by the reset path,
preventing stale handles from being sent to firmware.
4. A reset_rwsem serializes handle invalidation against resource creation
to avoid races between the reset path and new resource allocation.
Patches 1-6 introduce per-ucontext tracking lists for each resource type.
Patch 7 implements the reset/resume notification mechanism with rwsem
serialization, mmap revocation, and IB event dispatch.
Patch 8 adds INVALID_MANA_HANDLE checks in destroy callbacks.
Tested with DPDK testpmd on Azure VM (linux-next-20260306) — confirmed
IB_EVENT_PORT_ERR (type=10) and IB_EVENT_PORT_ACTIVE (type=9) are delivered
to userspace during service reset, and testpmd tears down cleanly afterwards.
Long Li (8):
RDMA/mana_ib: Track ucontext per device
RDMA/mana_ib: Track PD per ucontext
RDMA/mana_ib: Track CQ per ucontext
RDMA/mana_ib: Track WQ per ucontext
RDMA/mana_ib: Track QP per ucontext
RDMA/mana_ib: Track MR per ucontext
RDMA/mana_ib: Notify service reset events to RDMA devices
RDMA/mana_ib: Skip firmware commands for invalidated handles
drivers/infiniband/hw/mana/cq.c | 44 +++++--
drivers/infiniband/hw/mana/device.c | 105 ++++++++++++++++++
drivers/infiniband/hw/mana/main.c | 56 +++++++++-
drivers/infiniband/hw/mana/mana_ib.h | 19 ++++
drivers/infiniband/hw/mana/mr.c | 33 +++++-
drivers/infiniband/hw/mana/qp.c | 61 +++++++---
drivers/infiniband/hw/mana/wq.c | 24 ++++
drivers/net/ethernet/microsoft/mana/mana_en.c | 14 ++-
include/net/mana/gdma.h | 6 +
9 files changed, 331 insertions(+), 31 deletions(-)
--
2.43.0
^ permalink raw reply
* [PATCH rdma-next 1/8] RDMA/mana_ib: Track ucontext per device
From: Long Li @ 2026-03-07 1:47 UTC (permalink / raw)
To: Long Li, Konstantin Taranov, Jakub Kicinski, David S . Miller,
Paolo Abeni, Eric Dumazet, Andrew Lunn, Jason Gunthorpe,
Leon Romanovsky, Haiyang Zhang, K . Y . Srinivasan, Wei Liu,
Dexuan Cui
Cc: Simon Horman, netdev, linux-rdma, linux-hyperv, linux-kernel
In-Reply-To: <20260307014723.556523-1-longli@microsoft.com>
Add per-device tracking of ucontext objects. Each ucontext is added
to the device's ucontext_list on allocation and removed on deallocation.
A mutex protects the list and a per-ucontext lock protects resource
lists that will be added in subsequent patches.
This enables iterating over all active ucontexts during service reset
cleanup.
Signed-off-by: Long Li <longli@microsoft.com>
---
drivers/infiniband/hw/mana/device.c | 2 ++
drivers/infiniband/hw/mana/main.c | 10 ++++++++++
drivers/infiniband/hw/mana/mana_ib.h | 6 ++++++
3 files changed, 18 insertions(+)
diff --git a/drivers/infiniband/hw/mana/device.c b/drivers/infiniband/hw/mana/device.c
index ccc2279ca63c..149e8d4d5b8e 100644
--- a/drivers/infiniband/hw/mana/device.c
+++ b/drivers/infiniband/hw/mana/device.c
@@ -132,6 +132,8 @@ static int mana_ib_probe(struct auxiliary_device *adev,
dev->ib_dev.dev.parent = gc->dev;
dev->gdma_dev = mdev;
xa_init_flags(&dev->qp_table_wq, XA_FLAGS_LOCK_IRQ);
+ mutex_init(&dev->ucontext_lock);
+ INIT_LIST_HEAD(&dev->ucontext_list);
if (mana_ib_is_rnic(dev)) {
dev->ib_dev.phys_port_cnt = 1;
diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c
index 8d99cd00f002..fc28bdafcfd6 100644
--- a/drivers/infiniband/hw/mana/main.c
+++ b/drivers/infiniband/hw/mana/main.c
@@ -221,6 +221,12 @@ int mana_ib_alloc_ucontext(struct ib_ucontext *ibcontext,
ucontext->doorbell = doorbell_page;
+ mutex_init(&ucontext->lock);
+
+ mutex_lock(&mdev->ucontext_lock);
+ list_add_tail(&ucontext->dev_list, &mdev->ucontext_list);
+ mutex_unlock(&mdev->ucontext_lock);
+
return 0;
}
@@ -236,6 +242,10 @@ void mana_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
mdev = container_of(ibdev, struct mana_ib_dev, ib_dev);
gc = mdev_to_gc(mdev);
+ mutex_lock(&mdev->ucontext_lock);
+ list_del_init(&mana_ucontext->dev_list);
+ mutex_unlock(&mdev->ucontext_lock);
+
ret = mana_gd_destroy_doorbell_page(gc, mana_ucontext->doorbell);
if (ret)
ibdev_dbg(ibdev, "Failed to destroy doorbell page %d\n", ret);
diff --git a/drivers/infiniband/hw/mana/mana_ib.h b/drivers/infiniband/hw/mana/mana_ib.h
index a7c8c0fd7019..c7e333d3e9d8 100644
--- a/drivers/infiniband/hw/mana/mana_ib.h
+++ b/drivers/infiniband/hw/mana/mana_ib.h
@@ -83,6 +83,9 @@ struct mana_ib_dev {
struct dma_pool *av_pool;
netdevice_tracker dev_tracker;
struct notifier_block nb;
+ /* Protects ucontext_list */
+ struct mutex ucontext_lock;
+ struct list_head ucontext_list;
};
struct mana_ib_wq {
@@ -197,6 +200,9 @@ struct mana_ib_qp {
struct mana_ib_ucontext {
struct ib_ucontext ibucontext;
u32 doorbell;
+ struct list_head dev_list;
+ /* Protects resource lists below */
+ struct mutex lock;
};
struct mana_ib_rwq_ind_table {
--
2.43.0
^ permalink raw reply related
* [PATCH rdma-next 2/8] RDMA/mana_ib: Track PD per ucontext
From: Long Li @ 2026-03-07 1:47 UTC (permalink / raw)
To: Long Li, Konstantin Taranov, Jakub Kicinski, David S . Miller,
Paolo Abeni, Eric Dumazet, Andrew Lunn, Jason Gunthorpe,
Leon Romanovsky, Haiyang Zhang, K . Y . Srinivasan, Wei Liu,
Dexuan Cui
Cc: Simon Horman, netdev, linux-rdma, linux-hyperv, linux-kernel
In-Reply-To: <20260307014723.556523-1-longli@microsoft.com>
Add per-ucontext list tracking for PD objects. Each PD is added to
the ucontext's pd_list on creation and removed on destruction. This
enables iterating over all PDs belonging to a ucontext, which will
be needed for service reset cleanup.
Signed-off-by: Long Li <longli@microsoft.com>
---
drivers/infiniband/hw/mana/main.c | 21 +++++++++++++++++++++
drivers/infiniband/hw/mana/mana_ib.h | 2 ++
2 files changed, 23 insertions(+)
diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c
index fc28bdafcfd6..62d89ca06ba1 100644
--- a/drivers/infiniband/hw/mana/main.c
+++ b/drivers/infiniband/hw/mana/main.c
@@ -72,6 +72,7 @@ int mana_ib_alloc_pd(struct ib_pd *ibpd, struct ib_udata *udata)
struct ib_device *ibdev = ibpd->device;
struct gdma_create_pd_resp resp = {};
struct gdma_create_pd_req req = {};
+ struct mana_ib_ucontext *mana_ucontext;
enum gdma_pd_flags flags = 0;
struct mana_ib_dev *dev;
struct gdma_context *gc;
@@ -107,6 +108,16 @@ int mana_ib_alloc_pd(struct ib_pd *ibpd, struct ib_udata *udata)
mutex_init(&pd->vport_mutex);
pd->vport_use_count = 0;
+
+ INIT_LIST_HEAD(&pd->ucontext_list);
+ if (udata) {
+ mana_ucontext = rdma_udata_to_drv_context(udata,
+ struct mana_ib_ucontext, ibucontext);
+ mutex_lock(&mana_ucontext->lock);
+ list_add_tail(&pd->ucontext_list, &mana_ucontext->pd_list);
+ mutex_unlock(&mana_ucontext->lock);
+ }
+
return 0;
}
@@ -123,6 +134,15 @@ int mana_ib_dealloc_pd(struct ib_pd *ibpd, struct ib_udata *udata)
dev = container_of(ibdev, struct mana_ib_dev, ib_dev);
gc = mdev_to_gc(dev);
+ if (udata) {
+ struct mana_ib_ucontext *mana_ucontext =
+ rdma_udata_to_drv_context(udata,
+ struct mana_ib_ucontext, ibucontext);
+ mutex_lock(&mana_ucontext->lock);
+ list_del_init(&pd->ucontext_list);
+ mutex_unlock(&mana_ucontext->lock);
+ }
+
mana_gd_init_req_hdr(&req.hdr, GDMA_DESTROY_PD, sizeof(req),
sizeof(resp));
@@ -222,6 +242,7 @@ int mana_ib_alloc_ucontext(struct ib_ucontext *ibcontext,
ucontext->doorbell = doorbell_page;
mutex_init(&ucontext->lock);
+ INIT_LIST_HEAD(&ucontext->pd_list);
mutex_lock(&mdev->ucontext_lock);
list_add_tail(&ucontext->dev_list, &mdev->ucontext_list);
diff --git a/drivers/infiniband/hw/mana/mana_ib.h b/drivers/infiniband/hw/mana/mana_ib.h
index c7e333d3e9d8..6dba08bccc18 100644
--- a/drivers/infiniband/hw/mana/mana_ib.h
+++ b/drivers/infiniband/hw/mana/mana_ib.h
@@ -107,6 +107,7 @@ struct mana_ib_pd {
bool tx_shortform_allowed;
u32 tx_vp_offset;
+ struct list_head ucontext_list;
};
struct mana_ib_av {
@@ -203,6 +204,7 @@ struct mana_ib_ucontext {
struct list_head dev_list;
/* Protects resource lists below */
struct mutex lock;
+ struct list_head pd_list;
};
struct mana_ib_rwq_ind_table {
--
2.43.0
^ permalink raw reply related
* [PATCH rdma-next 3/8] RDMA/mana_ib: Track CQ per ucontext
From: Long Li @ 2026-03-07 1:47 UTC (permalink / raw)
To: Long Li, Konstantin Taranov, Jakub Kicinski, David S . Miller,
Paolo Abeni, Eric Dumazet, Andrew Lunn, Jason Gunthorpe,
Leon Romanovsky, Haiyang Zhang, K . Y . Srinivasan, Wei Liu,
Dexuan Cui
Cc: Simon Horman, netdev, linux-rdma, linux-hyperv, linux-kernel
In-Reply-To: <20260307014723.556523-1-longli@microsoft.com>
Add per-ucontext list tracking for CQ objects. Each CQ is added to
the ucontext's cq_list on creation and removed on destruction. This
enables iterating over all CQs belonging to a ucontext for service
reset cleanup.
Signed-off-by: Long Li <longli@microsoft.com>
---
drivers/infiniband/hw/mana/cq.c | 19 +++++++++++++++++++
drivers/infiniband/hw/mana/main.c | 1 +
drivers/infiniband/hw/mana/mana_ib.h | 2 ++
3 files changed, 22 insertions(+)
diff --git a/drivers/infiniband/hw/mana/cq.c b/drivers/infiniband/hw/mana/cq.c
index b2749f971cd0..89cf60987ff5 100644
--- a/drivers/infiniband/hw/mana/cq.c
+++ b/drivers/infiniband/hw/mana/cq.c
@@ -95,6 +95,16 @@ int mana_ib_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
INIT_LIST_HEAD(&cq->list_send_qp);
INIT_LIST_HEAD(&cq->list_recv_qp);
+ INIT_LIST_HEAD(&cq->ucontext_list);
+ if (udata) {
+ struct mana_ib_ucontext *mana_ucontext =
+ rdma_udata_to_drv_context(udata,
+ struct mana_ib_ucontext, ibucontext);
+ mutex_lock(&mana_ucontext->lock);
+ list_add_tail(&cq->ucontext_list, &mana_ucontext->cq_list);
+ mutex_unlock(&mana_ucontext->lock);
+ }
+
return 0;
err_remove_cq_cb:
@@ -115,6 +125,15 @@ int mana_ib_destroy_cq(struct ib_cq *ibcq, struct ib_udata *udata)
mdev = container_of(ibdev, struct mana_ib_dev, ib_dev);
+ if (udata) {
+ struct mana_ib_ucontext *mana_ucontext =
+ rdma_udata_to_drv_context(udata,
+ struct mana_ib_ucontext, ibucontext);
+ mutex_lock(&mana_ucontext->lock);
+ list_del_init(&cq->ucontext_list);
+ mutex_unlock(&mana_ucontext->lock);
+ }
+
mana_ib_remove_cq_cb(mdev, cq);
/* Ignore return code as there is not much we can do about it.
diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c
index 62d89ca06ba1..214c1d4e1548 100644
--- a/drivers/infiniband/hw/mana/main.c
+++ b/drivers/infiniband/hw/mana/main.c
@@ -243,6 +243,7 @@ int mana_ib_alloc_ucontext(struct ib_ucontext *ibcontext,
mutex_init(&ucontext->lock);
INIT_LIST_HEAD(&ucontext->pd_list);
+ INIT_LIST_HEAD(&ucontext->cq_list);
mutex_lock(&mdev->ucontext_lock);
list_add_tail(&ucontext->dev_list, &mdev->ucontext_list);
diff --git a/drivers/infiniband/hw/mana/mana_ib.h b/drivers/infiniband/hw/mana/mana_ib.h
index 6dba08bccc18..8d3edf7ba335 100644
--- a/drivers/infiniband/hw/mana/mana_ib.h
+++ b/drivers/infiniband/hw/mana/mana_ib.h
@@ -150,6 +150,7 @@ struct mana_ib_cq {
int cqe;
u32 comp_vector;
mana_handle_t cq_handle;
+ struct list_head ucontext_list;
};
enum mana_rc_queue_type {
@@ -205,6 +206,7 @@ struct mana_ib_ucontext {
/* Protects resource lists below */
struct mutex lock;
struct list_head pd_list;
+ struct list_head cq_list;
};
struct mana_ib_rwq_ind_table {
--
2.43.0
^ permalink raw reply related
* [PATCH rdma-next 4/8] RDMA/mana_ib: Track WQ per ucontext
From: Long Li @ 2026-03-07 1:47 UTC (permalink / raw)
To: Long Li, Konstantin Taranov, Jakub Kicinski, David S . Miller,
Paolo Abeni, Eric Dumazet, Andrew Lunn, Jason Gunthorpe,
Leon Romanovsky, Haiyang Zhang, K . Y . Srinivasan, Wei Liu,
Dexuan Cui
Cc: Simon Horman, netdev, linux-rdma, linux-hyperv, linux-kernel
In-Reply-To: <20260307014723.556523-1-longli@microsoft.com>
Add per-ucontext list tracking for WQ objects. Each WQ is added to
the ucontext's wq_list on creation and removed on destruction. This
enables iterating over all WQs belonging to a ucontext for service
reset cleanup.
Signed-off-by: Long Li <longli@microsoft.com>
---
drivers/infiniband/hw/mana/main.c | 1 +
drivers/infiniband/hw/mana/mana_ib.h | 2 ++
drivers/infiniband/hw/mana/wq.c | 20 ++++++++++++++++++++
3 files changed, 23 insertions(+)
diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c
index 214c1d4e1548..e6da5c8400f4 100644
--- a/drivers/infiniband/hw/mana/main.c
+++ b/drivers/infiniband/hw/mana/main.c
@@ -244,6 +244,7 @@ int mana_ib_alloc_ucontext(struct ib_ucontext *ibcontext,
mutex_init(&ucontext->lock);
INIT_LIST_HEAD(&ucontext->pd_list);
INIT_LIST_HEAD(&ucontext->cq_list);
+ INIT_LIST_HEAD(&ucontext->wq_list);
mutex_lock(&mdev->ucontext_lock);
list_add_tail(&ucontext->dev_list, &mdev->ucontext_list);
diff --git a/drivers/infiniband/hw/mana/mana_ib.h b/drivers/infiniband/hw/mana/mana_ib.h
index 8d3edf7ba335..96b5a13470ae 100644
--- a/drivers/infiniband/hw/mana/mana_ib.h
+++ b/drivers/infiniband/hw/mana/mana_ib.h
@@ -94,6 +94,7 @@ struct mana_ib_wq {
int wqe;
u32 wq_buf_size;
mana_handle_t rx_object;
+ struct list_head ucontext_list;
};
struct mana_ib_pd {
@@ -207,6 +208,7 @@ struct mana_ib_ucontext {
struct mutex lock;
struct list_head pd_list;
struct list_head cq_list;
+ struct list_head wq_list;
};
struct mana_ib_rwq_ind_table {
diff --git a/drivers/infiniband/hw/mana/wq.c b/drivers/infiniband/hw/mana/wq.c
index 6206244f762e..1af9869933aa 100644
--- a/drivers/infiniband/hw/mana/wq.c
+++ b/drivers/infiniband/hw/mana/wq.c
@@ -41,6 +41,17 @@ struct ib_wq *mana_ib_create_wq(struct ib_pd *pd,
wq->wqe = init_attr->max_wr;
wq->wq_buf_size = ucmd.wq_buf_size;
wq->rx_object = INVALID_MANA_HANDLE;
+
+ INIT_LIST_HEAD(&wq->ucontext_list);
+ if (udata) {
+ struct mana_ib_ucontext *mana_ucontext =
+ rdma_udata_to_drv_context(udata,
+ struct mana_ib_ucontext, ibucontext);
+ mutex_lock(&mana_ucontext->lock);
+ list_add_tail(&wq->ucontext_list, &mana_ucontext->wq_list);
+ mutex_unlock(&mana_ucontext->lock);
+ }
+
return &wq->ibwq;
err_free_wq:
@@ -64,6 +75,15 @@ int mana_ib_destroy_wq(struct ib_wq *ibwq, struct ib_udata *udata)
mdev = container_of(ib_dev, struct mana_ib_dev, ib_dev);
+ if (udata) {
+ struct mana_ib_ucontext *mana_ucontext =
+ rdma_udata_to_drv_context(udata,
+ struct mana_ib_ucontext, ibucontext);
+ mutex_lock(&mana_ucontext->lock);
+ list_del_init(&wq->ucontext_list);
+ mutex_unlock(&mana_ucontext->lock);
+ }
+
mana_ib_destroy_queue(mdev, &wq->queue);
kfree(wq);
--
2.43.0
^ permalink raw reply related
* [PATCH rdma-next 5/8] RDMA/mana_ib: Track QP per ucontext
From: Long Li @ 2026-03-07 1:47 UTC (permalink / raw)
To: Long Li, Konstantin Taranov, Jakub Kicinski, David S . Miller,
Paolo Abeni, Eric Dumazet, Andrew Lunn, Jason Gunthorpe,
Leon Romanovsky, Haiyang Zhang, K . Y . Srinivasan, Wei Liu,
Dexuan Cui
Cc: Simon Horman, netdev, linux-rdma, linux-hyperv, linux-kernel
In-Reply-To: <20260307014723.556523-1-longli@microsoft.com>
Add per-ucontext list tracking for QP objects. Only RAW_PACKET QPs
are tracked since they persist across reset events. RC, UD and GSI
QPs are removed and re-added during reset by IB core and do not
need tracking.
Signed-off-by: Long Li <longli@microsoft.com>
---
drivers/infiniband/hw/mana/main.c | 1 +
drivers/infiniband/hw/mana/mana_ib.h | 2 ++
drivers/infiniband/hw/mana/qp.c | 47 ++++++++++++++++++++++------
3 files changed, 40 insertions(+), 10 deletions(-)
diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c
index e6da5c8400f4..c6a859628ba3 100644
--- a/drivers/infiniband/hw/mana/main.c
+++ b/drivers/infiniband/hw/mana/main.c
@@ -244,6 +244,7 @@ int mana_ib_alloc_ucontext(struct ib_ucontext *ibcontext,
mutex_init(&ucontext->lock);
INIT_LIST_HEAD(&ucontext->pd_list);
INIT_LIST_HEAD(&ucontext->cq_list);
+ INIT_LIST_HEAD(&ucontext->qp_list);
INIT_LIST_HEAD(&ucontext->wq_list);
mutex_lock(&mdev->ucontext_lock);
diff --git a/drivers/infiniband/hw/mana/mana_ib.h b/drivers/infiniband/hw/mana/mana_ib.h
index 96b5a13470ae..9d90fda2c830 100644
--- a/drivers/infiniband/hw/mana/mana_ib.h
+++ b/drivers/infiniband/hw/mana/mana_ib.h
@@ -198,6 +198,7 @@ struct mana_ib_qp {
refcount_t refcount;
struct completion free;
+ struct list_head ucontext_list;
};
struct mana_ib_ucontext {
@@ -208,6 +209,7 @@ struct mana_ib_ucontext {
struct mutex lock;
struct list_head pd_list;
struct list_head cq_list;
+ struct list_head qp_list;
struct list_head wq_list;
};
diff --git a/drivers/infiniband/hw/mana/qp.c b/drivers/infiniband/hw/mana/qp.c
index 82f84f7ad37a..315bc54d8ae6 100644
--- a/drivers/infiniband/hw/mana/qp.c
+++ b/drivers/infiniband/hw/mana/qp.c
@@ -700,14 +700,31 @@ static int mana_ib_create_ud_qp(struct ib_qp *ibqp, struct ib_pd *ibpd,
int mana_ib_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *attr,
struct ib_udata *udata)
{
+ struct mana_ib_qp *qp = container_of(ibqp, struct mana_ib_qp, ibqp);
+ int err;
+
+ INIT_LIST_HEAD(&qp->ucontext_list);
+
switch (attr->qp_type) {
case IB_QPT_RAW_PACKET:
/* When rwq_ind_tbl is used, it's for creating WQs for RSS */
if (attr->rwq_ind_tbl)
- return mana_ib_create_qp_rss(ibqp, ibqp->pd, attr,
- udata);
+ err = mana_ib_create_qp_rss(ibqp, ibqp->pd, attr,
+ udata);
+ else
+ err = mana_ib_create_qp_raw(ibqp, ibqp->pd, attr,
+ udata);
+
+ if (!err && udata) {
+ struct mana_ib_ucontext *mana_ucontext =
+ rdma_udata_to_drv_context(udata,
+ struct mana_ib_ucontext, ibucontext);
+ mutex_lock(&mana_ucontext->lock);
+ list_add_tail(&qp->ucontext_list, &mana_ucontext->qp_list);
+ mutex_unlock(&mana_ucontext->lock);
+ }
- return mana_ib_create_qp_raw(ibqp, ibqp->pd, attr, udata);
+ return err;
case IB_QPT_RC:
return mana_ib_create_rc_qp(ibqp, ibqp->pd, attr, udata);
case IB_QPT_UD:
@@ -716,9 +733,8 @@ int mana_ib_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *attr,
default:
ibdev_dbg(ibqp->device, "Creating QP type %u not supported\n",
attr->qp_type);
+ return -EINVAL;
}
-
- return -EINVAL;
}
static int mana_ib_gd_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
@@ -898,14 +914,26 @@ static int mana_ib_destroy_ud_qp(struct mana_ib_qp *qp, struct ib_udata *udata)
int mana_ib_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata)
{
struct mana_ib_qp *qp = container_of(ibqp, struct mana_ib_qp, ibqp);
+ int ret = -ENOENT;
switch (ibqp->qp_type) {
case IB_QPT_RAW_PACKET:
+ if (udata) {
+ struct mana_ib_ucontext *mana_ucontext =
+ rdma_udata_to_drv_context(udata,
+ struct mana_ib_ucontext, ibucontext);
+ mutex_lock(&mana_ucontext->lock);
+ list_del_init(&qp->ucontext_list);
+ mutex_unlock(&mana_ucontext->lock);
+ }
+
if (ibqp->rwq_ind_tbl)
- return mana_ib_destroy_qp_rss(qp, ibqp->rwq_ind_tbl,
- udata);
+ ret = mana_ib_destroy_qp_rss(qp, ibqp->rwq_ind_tbl,
+ udata);
+ else
+ ret = mana_ib_destroy_qp_raw(qp, udata);
- return mana_ib_destroy_qp_raw(qp, udata);
+ return ret;
case IB_QPT_RC:
return mana_ib_destroy_rc_qp(qp, udata);
case IB_QPT_UD:
@@ -914,7 +942,6 @@ int mana_ib_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata)
default:
ibdev_dbg(ibqp->device, "Unexpected QP type %u\n",
ibqp->qp_type);
+ return ret;
}
-
- return -ENOENT;
}
--
2.43.0
^ permalink raw reply related
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox