crypto: qat - add resubmit logic for decompression

The acomp API allows to send requests with a NULL destination buffer. In
this case, the algorithm implementation needs to allocate the
destination scatter list, perform the operation and return the buffer to
the user. For decompression, data is likely to expand and be bigger than
the allocated buffer.

This implements a re-submission mechanism for decompression requests
that is triggered if the destination buffer, allocated by the driver,
is not sufficiently big to store the output from decompression.

If an overflow is detected when processing the callback for a
decompression request with a NULL destination buffer, a workqueue is
scheduled. This allocates a new scatter list of size CRYPTO_ACOMP_DST_MAX,
now 128KB, creates a new firmware scatter list and resubmits the job to
the hardware accelerator.
Suggested-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Giovanni Cabiddu <giovanni.cabiddu@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

drivers/crypto/qat/qat_common/qat_bl.c

@@ -222,3 +222,162 @@ int qat_bl_sgl_to_bufl(struct adf_accel_dev *accel_dev,
 				    extra_dst_buff, sz_extra_dst_buff,
 				    flags);
 }
+
+static void qat_bl_sgl_unmap(struct adf_accel_dev *accel_dev,
+			     struct qat_alg_buf_list *bl)
+{
+	struct device *dev = &GET_DEV(accel_dev);
+	int n = bl->num_bufs;
+	int i;
+
+	for (i = 0; i < n; i++)
+		if (!dma_mapping_error(dev, bl->bufers[i].addr))
+			dma_unmap_single(dev, bl->bufers[i].addr,
+					 bl->bufers[i].len, DMA_FROM_DEVICE);
+}
+
+static int qat_bl_sgl_map(struct adf_accel_dev *accel_dev,
+			  struct scatterlist *sgl,
+			  struct qat_alg_buf_list **bl)
+{
+	struct device *dev = &GET_DEV(accel_dev);
+	struct qat_alg_buf_list *bufl;
+	int node = dev_to_node(dev);
+	struct scatterlist *sg;
+	int n, i, sg_nctr;
+	size_t sz;
+
+	n = sg_nents(sgl);
+	sz = struct_size(bufl, bufers, n);
+	bufl = kzalloc_node(sz, GFP_KERNEL, node);
+	if (unlikely(!bufl))
+		return -ENOMEM;
+
+	for (i = 0; i < n; i++)
+		bufl->bufers[i].addr = DMA_MAPPING_ERROR;
+
+	sg_nctr = 0;
+	for_each_sg(sgl, sg, n, i) {
+		int y = sg_nctr;
+
+		if (!sg->length)
+			continue;
+
+		bufl->bufers[y].addr = dma_map_single(dev, sg_virt(sg),
+						      sg->length,
+						      DMA_FROM_DEVICE);
+		bufl->bufers[y].len = sg->length;
+		if (unlikely(dma_mapping_error(dev, bufl->bufers[y].addr)))
+			goto err_map;
+		sg_nctr++;
+	}
+	bufl->num_bufs = sg_nctr;
+	bufl->num_mapped_bufs = sg_nctr;
+
+	*bl = bufl;
+
+	return 0;
+
+err_map:
+	for (i = 0; i < n; i++)
+		if (!dma_mapping_error(dev, bufl->bufers[i].addr))
+			dma_unmap_single(dev, bufl->bufers[i].addr,
+					 bufl->bufers[i].len,
+					 DMA_FROM_DEVICE);
+	kfree(bufl);
+	*bl = NULL;
+
+	return -ENOMEM;
+}
+
+static void qat_bl_sgl_free_unmap(struct adf_accel_dev *accel_dev,
+				  struct scatterlist *sgl,
+				  struct qat_alg_buf_list *bl,
+				  bool free_bl)
+{
+	if (bl) {
+		qat_bl_sgl_unmap(accel_dev, bl);
+
+		if (free_bl)
+			kfree(bl);
+	}
+	if (sgl)
+		sgl_free(sgl);
+}
+
+static int qat_bl_sgl_alloc_map(struct adf_accel_dev *accel_dev,
+				struct scatterlist **sgl,
+				struct qat_alg_buf_list **bl,
+				unsigned int dlen,
+				gfp_t gfp)
+{
+	struct scatterlist *dst;
+	int ret;
+
+	dst = sgl_alloc(dlen, gfp, NULL);
+	if (!dst) {
+		dev_err(&GET_DEV(accel_dev), "sg_alloc failed\n");
+		return -ENOMEM;
+	}
+
+	ret = qat_bl_sgl_map(accel_dev, dst, bl);
+	if (ret)
+		goto err;
+
+	*sgl = dst;
+
+	return 0;
+
+err:
+	sgl_free(dst);
+	*sgl = NULL;
+	return ret;
+}
+
+int qat_bl_realloc_map_new_dst(struct adf_accel_dev *accel_dev,
+			       struct scatterlist **sg,
+			       unsigned int dlen,
+			       struct qat_request_buffs *qat_bufs,
+			       gfp_t gfp)
+{
+	struct device *dev = &GET_DEV(accel_dev);
+	dma_addr_t new_blp = DMA_MAPPING_ERROR;
+	struct qat_alg_buf_list *new_bl;
+	struct scatterlist *new_sg;
+	size_t new_bl_size;
+	int ret;
+
+	ret = qat_bl_sgl_alloc_map(accel_dev, &new_sg, &new_bl, dlen, gfp);
+	if (ret)
+		return ret;
+
+	new_bl_size = struct_size(new_bl, bufers, new_bl->num_bufs);
+
+	/* Map new firmware SGL descriptor */
+	new_blp = dma_map_single(dev, new_bl, new_bl_size, DMA_TO_DEVICE);
+	if (unlikely(dma_mapping_error(dev, new_blp)))
+		goto err;
+
+	/* Unmap old firmware SGL descriptor */
+	dma_unmap_single(dev, qat_bufs->bloutp, qat_bufs->sz_out, DMA_TO_DEVICE);
+
+	/* Free and unmap old scatterlist */
+	qat_bl_sgl_free_unmap(accel_dev, *sg, qat_bufs->blout,
+			      !qat_bufs->sgl_dst_valid);
+
+	qat_bufs->sgl_dst_valid = false;
+	qat_bufs->blout = new_bl;
+	qat_bufs->bloutp = new_blp;
+	qat_bufs->sz_out = new_bl_size;
+
+	*sg = new_sg;
+
+	return 0;
+err:
+	qat_bl_sgl_free_unmap(accel_dev, new_sg, new_bl, true);
+
+	if (!dma_mapping_error(dev, new_blp))
+		dma_unmap_single(dev, new_blp, new_bl_size, DMA_TO_DEVICE);
+
+	return -ENOMEM;
+}

drivers/crypto/qat/qat_common/qat_bl.h

@@ -58,4 +58,10 @@ static inline gfp_t qat_algs_alloc_flags(struct crypto_async_request *req)
 	return req->flags & CRYPTO_TFM_REQ_MAY_SLEEP ? GFP_KERNEL : GFP_ATOMIC;
 }
 
+int qat_bl_realloc_map_new_dst(struct adf_accel_dev *accel_dev,
+			       struct scatterlist **newd,
+			       unsigned int dlen,
+			       struct qat_request_buffs *qat_bufs,
+			       gfp_t gfp);
+
 #endif

drivers/crypto/qat/qat_common/qat_comp_algs.c

@@ -5,6 +5,7 @@
 #include <crypto/internal/acompress.h>
 #include <crypto/scatterwalk.h>
 #include <linux/dma-mapping.h>
+#include <linux/workqueue.h>
 #include "adf_accel_devices.h"
 #include "adf_common_drv.h"
 #include "qat_bl.h"
@@ -25,6 +26,11 @@ struct qat_compression_ctx {
 	struct qat_compression_instance *inst;
 };
 
+struct qat_dst {
+	bool is_null;
+	int resubmitted;
+};
+
 struct qat_compression_req {
 	u8 req[QAT_COMP_REQ_SIZE];
 	struct qat_compression_ctx *qat_compression_ctx;
@@ -33,6 +39,8 @@ struct qat_compression_req {
 	enum direction dir;
 	int actual_dlen;
 	struct qat_alg_req alg_req;
+	struct work_struct resubmit;
+	struct qat_dst dst;
 };
 
 static int qat_alg_send_dc_message(struct qat_compression_req *qat_req,
@@ -49,6 +57,46 @@ static int qat_alg_send_dc_message(struct qat_compression_req *qat_req,
 	return qat_alg_send_message(alg_req);
 }
 
+static void qat_comp_resubmit(struct work_struct *work)
+{
+	struct qat_compression_req *qat_req =
+		container_of(work, struct qat_compression_req, resubmit);
+	struct qat_compression_ctx *ctx = qat_req->qat_compression_ctx;
+	struct adf_accel_dev *accel_dev = ctx->inst->accel_dev;
+	struct qat_request_buffs *qat_bufs = &qat_req->buf;
+	struct qat_compression_instance *inst = ctx->inst;
+	struct acomp_req *areq = qat_req->acompress_req;
+	struct crypto_acomp *tfm = crypto_acomp_reqtfm(areq);
+	unsigned int dlen = CRYPTO_ACOMP_DST_MAX;
+	u8 *req = qat_req->req;
+	dma_addr_t dfbuf;
+	int ret;
+
+	areq->dlen = dlen;
+
+	dev_dbg(&GET_DEV(accel_dev), "[%s][%s] retry NULL dst request - dlen = %d\n",
+		crypto_tfm_alg_driver_name(crypto_acomp_tfm(tfm)),
+		qat_req->dir == COMPRESSION ? "comp" : "decomp", dlen);
+
+	ret = qat_bl_realloc_map_new_dst(accel_dev, &areq->dst, dlen, qat_bufs,
+					 qat_algs_alloc_flags(&areq->base));
+	if (ret)
+		goto err;
+
+	qat_req->dst.resubmitted = true;
+
+	dfbuf = qat_req->buf.bloutp;
+	qat_comp_override_dst(req, dfbuf, dlen);
+
+	ret = qat_alg_send_dc_message(qat_req, inst, &areq->base);
+	if (ret != -ENOSPC)
+		return;
+
+err:
+	qat_bl_free_bufl(accel_dev, qat_bufs);
+	areq->base.complete(&areq->base, ret);
+}
+
 static void qat_comp_generic_callback(struct qat_compression_req *qat_req,
 				      void *resp)
 {
@@ -80,6 +128,21 @@ static void qat_comp_generic_callback(struct qat_compression_req *qat_req,
 
 	areq->dlen = 0;
 
+	if (qat_req->dir == DECOMPRESSION && qat_req->dst.is_null) {
+		if (cmp_err == ERR_CODE_OVERFLOW_ERROR) {
+			if (qat_req->dst.resubmitted) {
+				dev_dbg(&GET_DEV(accel_dev),
+					"Output does not fit destination buffer\n");
+				res = -EOVERFLOW;
+				goto end;
+			}
+
+			INIT_WORK(&qat_req->resubmit, qat_comp_resubmit);
+			adf_misc_wq_queue_work(&qat_req->resubmit);
+			return;
+		}
+	}
+
 	if (unlikely(status != ICP_QAT_FW_COMN_STATUS_FLAG_OK))
 		goto end;
 
@@ -176,16 +239,23 @@ static int qat_comp_alg_compress_decompress(struct acomp_req *areq,
 	if (areq->dst && !dlen)
 		return -EINVAL;
 
+	qat_req->dst.is_null = false;
+
 	/* Handle acomp requests that require the allocation of a destination
 	 * buffer. The size of the destination buffer is double the source
 	 * buffer (rounded up to the size of a page) to fit the decompressed
 	 * output or an expansion on the data for compression.
 	 */
 	if (!areq->dst) {
+		qat_req->dst.is_null = true;
+
 		dlen = round_up(2 * slen, PAGE_SIZE);
 		areq->dst = sgl_alloc(dlen, f, NULL);
 		if (!areq->dst)
 			return -ENOMEM;
+
+		areq->dlen = dlen;
+		qat_req->dst.resubmitted = false;
 	}
 
 	if (dir == COMPRESSION) {

drivers/crypto/qat/qat_common/qat_comp_req.h

@@ -25,6 +25,16 @@ static inline void qat_comp_create_req(void *ctx, void *req, u64 src, u32 slen,
 	req_pars->out_buffer_sz = dlen;
 }
 
+static inline void qat_comp_override_dst(void *req, u64 dst, u32 dlen)
+{
+	struct icp_qat_fw_comp_req *fw_req = req;
+	struct icp_qat_fw_comp_req_params *req_pars = &fw_req->comp_pars;
+
+	fw_req->comn_mid.dest_data_addr = dst;
+	fw_req->comn_mid.dst_length = dlen;
+	req_pars->out_buffer_sz = dlen;
+}
+
 static inline void qat_comp_create_compression_req(void *ctx, void *req,
 						   u64 src, u32 slen,
 						   u64 dst, u32 dlen,