[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH v3] tests/qtest/libqos: add DMA support for writing and readi
From: |
Ani Sinha |
Subject: |
Re: [PATCH v3] tests/qtest/libqos: add DMA support for writing and reading fw_cfg files |
Date: |
Wed, 8 Jan 2025 18:47:25 +0530 |
> On 8 Jan 2025, at 6:38 PM, Daniel P. Berrangé <berrange@redhat.com> wrote:
>
> On Wed, Jan 08, 2025 at 06:27:50PM +0530, Ani Sinha wrote:
>> At present, the libqos/fw_cfg.c library does not support the modern DMA
>> interface which is required to write to the fw_cfg files. It only uses the IO
>> interface. Implement read and write methods based on DMA. This will enable
>> developers to write tests that writes to the fw_cfg file(s). The structure of
>> the code is taken from edk2 fw_cfg implementation. It has been tested by
>> writing a qtest that writes to a fw_cfg file. This test will be part of a
>> future patch series.
>>
>> Signed-off-by: Ani Sinha <anisinha@redhat.com>
>> ---
>> tests/qtest/libqos/fw_cfg.c | 204 ++++++++++++++++++++++++++++++++----
>> tests/qtest/libqos/fw_cfg.h | 5 +
>> 2 files changed, 186 insertions(+), 23 deletions(-)
>
>> +static bool
>> +find_pdir_entry(QFWCFG *fw_cfg, const char *filename,
>> + uint16_t *sel, uint32_t *size)
>> +{
>> + unsigned char *filesbuf = NULL;
>
> Use g_autofree here instead of later g_free.
OK will send just a refactoring patch with this change.
>
>> + uint32_t count;
>> + size_t dsize;
>> + FWCfgFile *pdir_entry;
>> + uint32_t i;
>> + bool found = false;
>> +
>> + *size = 0;
>> + *sel = 0;
>> +
>> + qfw_cfg_get(fw_cfg, FW_CFG_FILE_DIR, &count, sizeof(count));
>> + count = be32_to_cpu(count);
>> + dsize = sizeof(uint32_t) + count * sizeof(struct fw_cfg_file);
>> + filesbuf = g_malloc(dsize);
>> + g_assert(filesbuf);
>> + qfw_cfg_get(fw_cfg, FW_CFG_FILE_DIR, filesbuf, dsize);
>> + pdir_entry = (FWCfgFile *)(filesbuf + sizeof(uint32_t));
>
> I'm not familiar with fwcfg data format, but I'm wondering
> what the initial 'uint32_t' data field is that you're skipping
> over, and whether its value should be validated before this
> loop ?
This part I left as is from previous code. From
https://www.qemu.org/docs/master/specs/fw_cfg.html
struct FWCfgFiles { /* the entire file directory fw_cfg item */
uint32_t count; /* number of entries, in big-endian format */
struct FWCfgFile f[]; /* array of file entries, see below */
};
struct FWCfgFile { /* an individual file entry, 64 bytes total */
uint32_t size; /* size of referenced fw_cfg item, big-endian */
uint16_t select; /* selector key of fw_cfg item, big-endian */
uint16_t reserved;
char name[56]; /* fw_cfg item name, NUL-terminated ascii */
};
So the code first reads the count and then allocates ‘count' entries for
‘count' files.
>
>> + for (i = 0; i < count; ++i, ++pdir_entry) {
>> + if (!strcmp(pdir_entry->name, filename)) {
>> + *size = be32_to_cpu(pdir_entry->size);
>> + *sel = be16_to_cpu(pdir_entry->select);
>> + found = true;
>> + break;
>> + }
>> + }
>> +
>> + g_free(filesbuf);
>> + return found;
>> +}
>
>> +
>> /*
>> * The caller need check the return value. When the return value is
>> * nonzero, it means that some bytes have been transferred.
>> @@ -73,37 +168,100 @@ static void mm_fw_cfg_select(QFWCFG *fw_cfg, uint16_t
>> key)
>> * populated, it has received only a starting slice of the fw_cfg file.
>> */
>> size_t qfw_cfg_get_file(QFWCFG *fw_cfg, const char *filename,
>> - void *data, size_t buflen)
>> + void *data, size_t buflen)
>> {
>> - uint32_t count;
>> - uint32_t i;
>> - unsigned char *filesbuf = NULL;
>> - size_t dsize;
>> - FWCfgFile *pdir_entry;
>> size_t filesize = 0;
>> + uint32_t len;
>> + uint16_t sel;
>>
>> - qfw_cfg_get(fw_cfg, FW_CFG_FILE_DIR, &count, sizeof(count));
>> - count = be32_to_cpu(count);
>> - dsize = sizeof(uint32_t) + count * sizeof(struct fw_cfg_file);
>> - filesbuf = g_malloc(dsize);
>> - qfw_cfg_get(fw_cfg, FW_CFG_FILE_DIR, filesbuf, dsize);
>> - pdir_entry = (FWCfgFile *)(filesbuf + sizeof(uint32_t));
>> - for (i = 0; i < count; ++i, ++pdir_entry) {
>> - if (!strcmp(pdir_entry->name, filename)) {
>> - uint32_t len = be32_to_cpu(pdir_entry->size);
>> - uint16_t sel = be16_to_cpu(pdir_entry->select);
>> - filesize = len;
>> - if (len > buflen) {
>> - len = buflen;
>> - }
>> - qfw_cfg_get(fw_cfg, sel, data, len);
>> - break;
>> + if (find_pdir_entry(fw_cfg, filename, &sel, &len)) {
>> + filesize = len;
>> + if (len > buflen) {
>> + len = buflen;
>> }
>> + qfw_cfg_get(fw_cfg, sel, data, len);
>> }
>> - g_free(filesbuf);
>> +
>> return filesize;
>> }
>
> I'd recommend refactoring of existnig code, be a separate commit
> from the newly added functionality.
Yes will do that.
>
>>
>> +/*
>> + * The caller need check the return value. When the return value is
>> + * nonzero, it means that some bytes have been transferred.
>> + *
>> + * If the fw_cfg file in question is smaller than the allocated & passed-in
>> + * buffer, then the first len bytes were read.
>> + *
>> + * If the fw_cfg file in question is larger than the passed-in
>> + * buffer, then the return value explains how much was actually read.
>> + *
>> + * It is illegal to call this function if fw_cfg does not support DMA
>> + * interface. The caller should ensure that DMA is supported before
>> + * calling this function.
>> + *
>> + * Passed QOSState pointer qs must be initialized. qs->alloc must also be
>> + * properly initialized.
>> + */
>> +size_t qfw_cfg_read_file(QFWCFG *fw_cfg, QOSState *qs, const char *filename,
>> + void *data, size_t buflen)
>> +{
>> + uint32_t len = 0;
>> + uint16_t sel;
>> + uint32_t id;
>> +
>> + g_assert(qs);
>> + /* check if DMA is supported since we use DMA for read */
>> + id = qfw_cfg_get_u32(fw_cfg, FW_CFG_ID);
>> + g_assert(id & FW_CFG_VERSION_DMA);
>> +
>> + if (find_pdir_entry(fw_cfg, filename, &sel, &len)) {
>> + if (len > buflen) {
>> + len = buflen;
>> + }
>> + qfw_cfg_read_entry(fw_cfg, qs, sel, data, len);
>> + }
>> +
>> + return (size_t) len;
>
> The size_t cast is redundant, since we know sizeof(size_t)
> will be >= sizeof(uint32_t) on all platforms
>
>> +}
>> +
>> +/*
>> + * The caller need check the return value. When the return value is
>> + * nonzero, it means that some bytes have been transferred.
>> + *
>> + * If the fw_cfg file in question is smaller than the allocated & passed-in
>> + * buffer, then the buffer has been partially written.
>> + *
>> + * If the fw_cfg file in question is larger than the passed-in
>> + * buffer, then the return value explains how much was actually written.
>> + *
>> + * It is illegal to call this function if fw_cfg does not support DMA
>> + * interface. The caller should ensure that DMA is supported before
>> + * calling this function.
>> + *
>> + * Passed QOSState pointer qs must be initialized. qs->alloc must also be
>> + * properly initialized.
>> + */
>> +size_t qfw_cfg_write_file(QFWCFG *fw_cfg, QOSState *qs, const char
>> *filename,
>> + void *data, size_t buflen)
>> +{
>> + uint32_t len = 0;
>> + uint16_t sel;
>> + uint32_t id;
>> +
>> + g_assert(qs);
>> + /* write operation is only valid if DMA is supported */
>> + id = qfw_cfg_get_u32(fw_cfg, FW_CFG_ID);
>> + g_assert(id & FW_CFG_VERSION_DMA);
>> +
>> + if (find_pdir_entry(fw_cfg, filename, &sel, &len)) {
>> + if (len > buflen) {
>> + len = buflen;
>> + }
>> + qfw_cfg_write_entry(fw_cfg, qs, sel, data, len);
>> + }
>> + return (size_t) len;
>
> Another redundant cast
>
>
> With regards,
> Daniel
> --
> |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :|
> |: https://libvirt.org -o- https://fstop138.berrange.com :|
> |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|