@@ -686,7 +686,7 @@ static int qat_alg_sgl_to_bufl(struct qat_crypto_instance *inst,
blp = dma_map_single(dev, bufl, sz, DMA_TO_DEVICE);
if (unlikely(dma_mapping_error(dev, blp)))
- goto err;
+ goto err_in;
for_each_sg(sgl, sg, n, i) {
int y = sg_nctr;
@@ -699,7 +699,7 @@ static int qat_alg_sgl_to_bufl(struct qat_crypto_instance *inst,
DMA_BIDIRECTIONAL);
bufl->bufers[y].len = sg->length;
if (unlikely(dma_mapping_error(dev, bufl->bufers[y].addr)))
- goto err;
+ goto err_in;
sg_nctr++;
}
bufl->num_bufs = sg_nctr;
@@ -717,10 +717,10 @@ static int qat_alg_sgl_to_bufl(struct qat_crypto_instance *inst,
buflout = kzalloc_node(sz_out, GFP_ATOMIC,
dev_to_node(&GET_DEV(inst->accel_dev)));
if (unlikely(!buflout))
- goto err;
+ goto err_in;
bloutp = dma_map_single(dev, buflout, sz_out, DMA_TO_DEVICE);
if (unlikely(dma_mapping_error(dev, bloutp)))
- goto err;
+ goto err_out;
bufers = buflout->bufers;
for_each_sg(sglout, sg, n, i) {
int y = sg_nctr;
@@ -732,7 +732,7 @@ static int qat_alg_sgl_to_bufl(struct qat_crypto_instance *inst,
sg->length,
DMA_BIDIRECTIONAL);
if (unlikely(dma_mapping_error(dev, bufers[y].addr)))
- goto err;
+ goto err_out;
bufers[y].len = sg->length;
sg_nctr++;
}
@@ -747,9 +747,20 @@ static int qat_alg_sgl_to_bufl(struct qat_crypto_instance *inst,
qat_req->buf.sz_out = 0;
}
return 0;
-err:
- dev_err(dev, "Failed to map buf for dma\n");
- sg_nctr = 0;
+
+err_out:
+ n = sg_nents(sglout);
+ for (i = 0; i < n; i++)
+ if (!dma_mapping_error(dev, buflout->bufers[i].addr))
+ dma_unmap_single(dev, buflout->bufers[i].addr,
+ buflout->bufers[i].len,
+ DMA_BIDIRECTIONAL);
+ if (!dma_mapping_error(dev, bloutp))
+ dma_unmap_single(dev, bloutp, sz_out, DMA_TO_DEVICE);
+ kfree(buflout);
+
+err_in:
+ n = sg_nents(sgl);
for (i = 0; i < n; i++)
if (!dma_mapping_error(dev, bufl->bufers[i].addr))
dma_unmap_single(dev, bufl->bufers[i].addr,
@@ -759,17 +770,8 @@ static int qat_alg_sgl_to_bufl(struct qat_crypto_instance *inst,
if (!dma_mapping_error(dev, blp))
dma_unmap_single(dev, blp, sz, DMA_TO_DEVICE);
kfree(bufl);
- if (sgl != sglout && buflout) {
- n = sg_nents(sglout);
- for (i = 0; i < n; i++)
- if (!dma_mapping_error(dev, buflout->bufers[i].addr))
- dma_unmap_single(dev, buflout->bufers[i].addr,
- buflout->bufers[i].len,
- DMA_BIDIRECTIONAL);
- if (!dma_mapping_error(dev, bloutp))
- dma_unmap_single(dev, bloutp, sz_out, DMA_TO_DEVICE);
- kfree(buflout);
- }
+
+ dev_err(dev, "Failed to map buf for dma\n");
return -ENOMEM;
}
After commit 9e442aa6a753 ("x86: remove DMA_ERROR_CODE"), the inlining decisions in the qat driver changed slightly, introducing a new false-positive warning: drivers/crypto/qat/qat_common/qat_algs.c: In function 'qat_alg_sgl_to_bufl.isra.6': include/linux/dma-mapping.h:228:2: error: 'sz_out' may be used uninitialized in this function [-Werror=maybe-uninitialized] drivers/crypto/qat/qat_common/qat_algs.c:676:9: note: 'sz_out' was declared here The patch that introduced this is correct, so let's just avoid the warning in this driver by rearranging the unwinding after an error to make it more obvious to the compiler what is going on. The problem here is the 'if (unlikely(dma_mapping_error(dev, blp)))' check, in which the 'unlikely' causes gcc to forget what it knew about the state of the variables. Cleaning up the dma state in the reverse order it was created means we can simplify the logic so it doesn't have to know about that state, and also makes it easier to understand. Cc: Christoph Hellwig <hch@lst.de> Signed-off-by: Arnd Bergmann <arnd@arndb.de> --- drivers/crypto/qat/qat_common/qat_algs.c | 40 +++++++++++++++++--------------- 1 file changed, 21 insertions(+), 19 deletions(-) -- 2.9.0