* [PATCH] crypto: octeontx2: CN10K CPT to RNM workaround
@ 2022-01-25 18:26 Shijith Thotton
2022-01-25 18:26 ` [PATCH] crypto: octeontx2: disable DMA black hole on an DMA fault Shijith Thotton
` (2 more replies)
0 siblings, 3 replies; 6+ messages in thread
From: Shijith Thotton @ 2022-01-25 18:26 UTC (permalink / raw)
To: Arnaud Ebalard, Herbert Xu, Boris Brezillon
Cc: Srujana Challa, linux-crypto, jerinj, sgoutham, Shijith Thotton,
David S. Miller, Dan Carpenter, Jiapeng Chong, open list
From: Srujana Challa <schalla@marvell.com>
When software sets CPT_AF_CTL[RNM_REQ_EN]=1 and RNM in not producing
entropy(i.e., RNM_ENTROPY_STATUS[NORMAL_CNT] < 0x40), the first cycle of
the response may be lost due to a conditional clocking issue. Due to
this, the subsequent random number stream will be corrupted. So, this
patch adds support to ensure RNM_ENTROPY_STATUS[NORMAL_CNT] = 0x40
before writing CPT_AF_CTL[RNM_REQ_EN] = 1, as a workaround.
Signed-off-by: Srujana Challa <schalla@marvell.com>
Signed-off-by: Shijith Thotton <sthotton@marvell.com>
---
.../marvell/octeontx2/otx2_cptpf_ucode.c | 43 ++++++++++++++++++-
1 file changed, 42 insertions(+), 1 deletion(-)
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c b/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c
index 4c8ebdf671ca..addc760501a9 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c
@@ -1076,6 +1076,39 @@ static void delete_engine_grps(struct pci_dev *pdev,
delete_engine_group(&pdev->dev, &eng_grps->grp[i]);
}
+#define PCI_DEVID_CN10K_RNM 0xA098
+#define RNM_ENTROPY_STATUS 0x8
+
+static void rnm_to_cpt_errata_fixup(struct device *dev)
+{
+ struct pci_dev *pdev;
+ void __iomem *base;
+ int timeout = 5000;
+
+ pdev = pci_get_device(PCI_VENDOR_ID_CAVIUM, PCI_DEVID_CN10K_RNM, NULL);
+ if (!pdev)
+ return;
+
+ base = pci_ioremap_bar(pdev, 0);
+ if (!base)
+ goto put_pdev;
+
+ while ((readq(base + RNM_ENTROPY_STATUS) & 0x7F) != 0x40) {
+ cpu_relax();
+ udelay(1);
+ timeout--;
+ if (!timeout) {
+ dev_warn(dev, "RNM is not producing entropy\n");
+ break;
+ }
+ }
+
+ iounmap(base);
+
+put_pdev:
+ pci_dev_put(pdev);
+}
+
int otx2_cpt_get_eng_grp(struct otx2_cpt_eng_grps *eng_grps, int eng_type)
{
@@ -1189,9 +1222,17 @@ int otx2_cpt_create_eng_grps(struct otx2_cptpf_dev *cptpf,
if (is_dev_otx2(pdev))
goto unlock;
+
+ /*
+ * Ensure RNM_ENTROPY_STATUS[NORMAL_CNT] = 0x40 before writing
+ * CPT_AF_CTL[RNM_REQ_EN] = 1 as a workaround for HW errata.
+ */
+ rnm_to_cpt_errata_fixup(&pdev->dev);
+
/*
* Configure engine group mask to allow context prefetching
- * for the groups.
+ * for the groups and enable random number request, to enable
+ * CPT to request random numbers from RNM.
*/
otx2_cpt_write_af_reg(&cptpf->afpf_mbox, pdev, CPT_AF_CTL,
OTX2_CPT_ALL_ENG_GRPS_MASK << 3 | BIT_ULL(16),
--
2.25.1
^ permalink raw reply related [flat|nested] 6+ messages in thread
* [PATCH] crypto: octeontx2: disable DMA black hole on an DMA fault
2022-01-25 18:26 [PATCH] crypto: octeontx2: CN10K CPT to RNM workaround Shijith Thotton
@ 2022-01-25 18:26 ` Shijith Thotton
2022-02-05 4:30 ` Herbert Xu
2022-01-25 18:26 ` [PATCH] crypto: octeontx2: increase CPT HW instruction queue length Shijith Thotton
2022-02-05 4:30 ` [PATCH] crypto: octeontx2: CN10K CPT to RNM workaround Herbert Xu
2 siblings, 1 reply; 6+ messages in thread
From: Shijith Thotton @ 2022-01-25 18:26 UTC (permalink / raw)
To: Arnaud Ebalard, Herbert Xu, Boris Brezillon
Cc: Srujana Challa, linux-crypto, jerinj, sgoutham, Shijith Thotton,
David S. Miller, Linu Cherian, Geetha sowjanya, hariprasad,
Subbaraya Sundeep, Jakub Kicinski, Dan Carpenter, Jiapeng Chong,
open list, open list:MARVELL OCTEONTX2 RVU ADMIN FUNCTION DRIVER
From: Srujana Challa <schalla@marvell.com>
When CPT_AF_DIAG[FLT_DIS] = 0 and a CPT engine access to
LLC/DRAM encounters a fault/poison, a rare case may result
in unpredictable data being delivered to a CPT engine.
So, this patch adds code to set FLT_DIS as a workaround.
Signed-off-by: Srujana Challa <schalla@marvell.com>
Signed-off-by: Shijith Thotton <sthotton@marvell.com>
---
drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c | 13 +++++++++++++
drivers/net/ethernet/marvell/octeontx2/af/rvu_cpt.c | 1 +
2 files changed, 14 insertions(+)
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c b/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c
index 4c8ebdf671ca..e0b29cf504b9 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c
@@ -1111,6 +1111,7 @@ int otx2_cpt_create_eng_grps(struct otx2_cptpf_dev *cptpf,
struct otx2_cpt_engines engs[OTX2_CPT_MAX_ETYPES_PER_GRP] = { {0} };
struct pci_dev *pdev = cptpf->pdev;
struct fw_info_t fw_info;
+ u64 reg_val;
int ret = 0;
mutex_lock(&eng_grps->lock);
@@ -1203,6 +1204,18 @@ int otx2_cpt_create_eng_grps(struct otx2_cptpf_dev *cptpf,
*/
otx2_cpt_write_af_reg(&cptpf->afpf_mbox, pdev, CPT_AF_CTX_FLUSH_TIMER,
CTX_FLUSH_TIMER_CNT, BLKADDR_CPT0);
+
+ /*
+ * Set CPT_AF_DIAG[FLT_DIS], as a workaround for HW errata, when
+ * CPT_AF_DIAG[FLT_DIS] = 0 and a CPT engine access to LLC/DRAM
+ * encounters a fault/poison, a rare case may result in
+ * unpredictable data being delivered to a CPT engine.
+ */
+ otx2_cpt_read_af_reg(&cptpf->afpf_mbox, pdev, CPT_AF_DIAG, ®_val,
+ BLKADDR_CPT0);
+ otx2_cpt_write_af_reg(&cptpf->afpf_mbox, pdev, CPT_AF_DIAG,
+ reg_val | BIT_ULL(24), BLKADDR_CPT0);
+
mutex_unlock(&eng_grps->lock);
return 0;
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_cpt.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_cpt.c
index 45357deecabb..1f7c971e6757 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_cpt.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_cpt.c
@@ -606,6 +606,7 @@ static bool is_valid_offset(struct rvu *rvu, struct cpt_rd_wr_reg_msg *req)
} else if (!(req->hdr.pcifunc & RVU_PFVF_FUNC_MASK)) {
/* Registers that can be accessed from PF */
switch (offset) {
+ case CPT_AF_DIAG:
case CPT_AF_CTL:
case CPT_AF_PF_FUNC:
case CPT_AF_BLK_RST:
--
2.25.1
^ permalink raw reply related [flat|nested] 6+ messages in thread
* [PATCH] crypto: octeontx2: increase CPT HW instruction queue length
2022-01-25 18:26 [PATCH] crypto: octeontx2: CN10K CPT to RNM workaround Shijith Thotton
2022-01-25 18:26 ` [PATCH] crypto: octeontx2: disable DMA black hole on an DMA fault Shijith Thotton
@ 2022-01-25 18:26 ` Shijith Thotton
2022-02-05 4:30 ` Herbert Xu
2022-02-05 4:30 ` [PATCH] crypto: octeontx2: CN10K CPT to RNM workaround Herbert Xu
2 siblings, 1 reply; 6+ messages in thread
From: Shijith Thotton @ 2022-01-25 18:26 UTC (permalink / raw)
To: Arnaud Ebalard, Herbert Xu, Boris Brezillon
Cc: Srujana Challa, linux-crypto, jerinj, sgoutham, Shijith Thotton,
David S. Miller, open list
From: Srujana Challa <schalla@marvell.com>
LDWB is getting incorrectly used in HW when
CPT_AF_LF()_PTR_CTL[IQB_LDWB]=1 and CPT instruction queue has less than
320 free entries. So, increase HW instruction queue size by 320 and give
320 entries less for SW/NIX RX as a SW workaround.
Signed-off-by: Srujana Challa <schalla@marvell.com>
Signed-off-by: Shijith Thotton <sthotton@marvell.com>
---
drivers/crypto/marvell/octeontx2/otx2_cptlf.h | 19 +++++++++++++++----
1 file changed, 15 insertions(+), 4 deletions(-)
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptlf.h b/drivers/crypto/marvell/octeontx2/otx2_cptlf.h
index b691b6c1d5c4..4fcaf61a70e3 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptlf.h
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptlf.h
@@ -26,12 +26,22 @@
*/
#define OTX2_CPT_INST_QLEN_MSGS ((OTX2_CPT_SIZE_DIV40 - 1) * 40)
+/*
+ * LDWB is getting incorrectly used when IQB_LDWB = 1 and CPT instruction
+ * queue has less than 320 free entries. So, increase HW instruction queue
+ * size by 320 and give 320 entries less for SW/NIX RX as a workaround.
+ */
+#define OTX2_CPT_INST_QLEN_EXTRA_BYTES (320 * OTX2_CPT_INST_SIZE)
+#define OTX2_CPT_EXTRA_SIZE_DIV40 (320/40)
+
/* CPT instruction queue length in bytes */
-#define OTX2_CPT_INST_QLEN_BYTES (OTX2_CPT_SIZE_DIV40 * 40 * \
- OTX2_CPT_INST_SIZE)
+#define OTX2_CPT_INST_QLEN_BYTES \
+ ((OTX2_CPT_SIZE_DIV40 * 40 * OTX2_CPT_INST_SIZE) + \
+ OTX2_CPT_INST_QLEN_EXTRA_BYTES)
/* CPT instruction group queue length in bytes */
-#define OTX2_CPT_INST_GRP_QLEN_BYTES (OTX2_CPT_SIZE_DIV40 * 16)
+#define OTX2_CPT_INST_GRP_QLEN_BYTES \
+ ((OTX2_CPT_SIZE_DIV40 + OTX2_CPT_EXTRA_SIZE_DIV40) * 16)
/* CPT FC length in bytes */
#define OTX2_CPT_Q_FC_LEN 128
@@ -179,7 +189,8 @@ static inline void otx2_cptlf_do_set_iqueue_size(struct otx2_cptlf_info *lf)
{
union otx2_cptx_lf_q_size lf_q_size = { .u = 0x0 };
- lf_q_size.s.size_div40 = OTX2_CPT_SIZE_DIV40;
+ lf_q_size.s.size_div40 = OTX2_CPT_SIZE_DIV40 +
+ OTX2_CPT_EXTRA_SIZE_DIV40;
otx2_cpt_write64(lf->lfs->reg_base, BLKADDR_CPT0, lf->slot,
OTX2_CPT_LF_Q_SIZE, lf_q_size.u);
}
--
2.25.1
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH] crypto: octeontx2: CN10K CPT to RNM workaround
2022-01-25 18:26 [PATCH] crypto: octeontx2: CN10K CPT to RNM workaround Shijith Thotton
2022-01-25 18:26 ` [PATCH] crypto: octeontx2: disable DMA black hole on an DMA fault Shijith Thotton
2022-01-25 18:26 ` [PATCH] crypto: octeontx2: increase CPT HW instruction queue length Shijith Thotton
@ 2022-02-05 4:30 ` Herbert Xu
2 siblings, 0 replies; 6+ messages in thread
From: Herbert Xu @ 2022-02-05 4:30 UTC (permalink / raw)
To: Shijith Thotton
Cc: Arnaud Ebalard, Boris Brezillon, Srujana Challa, linux-crypto,
jerinj, sgoutham, David S. Miller, Dan Carpenter, Jiapeng Chong,
open list
On Tue, Jan 25, 2022 at 11:56:22PM +0530, Shijith Thotton wrote:
> From: Srujana Challa <schalla@marvell.com>
>
> When software sets CPT_AF_CTL[RNM_REQ_EN]=1 and RNM in not producing
> entropy(i.e., RNM_ENTROPY_STATUS[NORMAL_CNT] < 0x40), the first cycle of
> the response may be lost due to a conditional clocking issue. Due to
> this, the subsequent random number stream will be corrupted. So, this
> patch adds support to ensure RNM_ENTROPY_STATUS[NORMAL_CNT] = 0x40
> before writing CPT_AF_CTL[RNM_REQ_EN] = 1, as a workaround.
>
> Signed-off-by: Srujana Challa <schalla@marvell.com>
> Signed-off-by: Shijith Thotton <sthotton@marvell.com>
> ---
> .../marvell/octeontx2/otx2_cptpf_ucode.c | 43 ++++++++++++++++++-
> 1 file changed, 42 insertions(+), 1 deletion(-)
Patch applied. Thanks.
--
Email: Herbert Xu <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] crypto: octeontx2: disable DMA black hole on an DMA fault
2022-01-25 18:26 ` [PATCH] crypto: octeontx2: disable DMA black hole on an DMA fault Shijith Thotton
@ 2022-02-05 4:30 ` Herbert Xu
0 siblings, 0 replies; 6+ messages in thread
From: Herbert Xu @ 2022-02-05 4:30 UTC (permalink / raw)
To: Shijith Thotton
Cc: Arnaud Ebalard, Boris Brezillon, Srujana Challa, linux-crypto,
jerinj, sgoutham, David S. Miller, Linu Cherian, Geetha sowjanya,
hariprasad, Subbaraya Sundeep, Jakub Kicinski, Dan Carpenter,
Jiapeng Chong, open list,
open list:MARVELL OCTEONTX2 RVU ADMIN FUNCTION DRIVER
On Tue, Jan 25, 2022 at 11:56:23PM +0530, Shijith Thotton wrote:
> From: Srujana Challa <schalla@marvell.com>
>
> When CPT_AF_DIAG[FLT_DIS] = 0 and a CPT engine access to
> LLC/DRAM encounters a fault/poison, a rare case may result
> in unpredictable data being delivered to a CPT engine.
> So, this patch adds code to set FLT_DIS as a workaround.
>
> Signed-off-by: Srujana Challa <schalla@marvell.com>
> Signed-off-by: Shijith Thotton <sthotton@marvell.com>
> ---
> drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c | 13 +++++++++++++
> drivers/net/ethernet/marvell/octeontx2/af/rvu_cpt.c | 1 +
> 2 files changed, 14 insertions(+)
Patch applied. Thanks.
--
Email: Herbert Xu <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] crypto: octeontx2: increase CPT HW instruction queue length
2022-01-25 18:26 ` [PATCH] crypto: octeontx2: increase CPT HW instruction queue length Shijith Thotton
@ 2022-02-05 4:30 ` Herbert Xu
0 siblings, 0 replies; 6+ messages in thread
From: Herbert Xu @ 2022-02-05 4:30 UTC (permalink / raw)
To: Shijith Thotton
Cc: Arnaud Ebalard, Boris Brezillon, Srujana Challa, linux-crypto,
jerinj, sgoutham, David S. Miller, open list
On Tue, Jan 25, 2022 at 11:56:24PM +0530, Shijith Thotton wrote:
> From: Srujana Challa <schalla@marvell.com>
>
> LDWB is getting incorrectly used in HW when
> CPT_AF_LF()_PTR_CTL[IQB_LDWB]=1 and CPT instruction queue has less than
> 320 free entries. So, increase HW instruction queue size by 320 and give
> 320 entries less for SW/NIX RX as a SW workaround.
>
> Signed-off-by: Srujana Challa <schalla@marvell.com>
> Signed-off-by: Shijith Thotton <sthotton@marvell.com>
> ---
> drivers/crypto/marvell/octeontx2/otx2_cptlf.h | 19 +++++++++++++++----
> 1 file changed, 15 insertions(+), 4 deletions(-)
Patch applied. Thanks.
--
Email: Herbert Xu <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2022-02-05 4:30 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-01-25 18:26 [PATCH] crypto: octeontx2: CN10K CPT to RNM workaround Shijith Thotton
2022-01-25 18:26 ` [PATCH] crypto: octeontx2: disable DMA black hole on an DMA fault Shijith Thotton
2022-02-05 4:30 ` Herbert Xu
2022-01-25 18:26 ` [PATCH] crypto: octeontx2: increase CPT HW instruction queue length Shijith Thotton
2022-02-05 4:30 ` Herbert Xu
2022-02-05 4:30 ` [PATCH] crypto: octeontx2: CN10K CPT to RNM workaround Herbert Xu
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).